Embryo-fetal safety evaluation of ondansetron in rats.

BACKGROUND: Ondansetron is a 5HT3 receptor antagonist, used to mitigate the effects of nausea and vomiting after chemotherapy or surgery. Since nausea and vomiting are common experiences during the first trimester of pregnancy, this antiemetic has been the main drug used during this period. METHODS: To evaluate the effects of ondansetron on the embryo-fetal development, which are still very contradictory, pregnant rats were exposed to therapeutic doses of ondansetron (1.7 or 2.5 mg/kg) daily, from gestational day (GD) 6 to 15. RESULTS: No clinical signs of toxicity were observed in dams during the treatment. Although the hemato-biochemical parameters were similar among the groups, histological changes, as well as a reduction in the weight of kidney were found in the treated dams. After fetal examination, no visceral and skeletal abnormalities were observed in treated fetuses. CONCLUSION: In conclusion, therapeutic doses of ondansetron have low teratogenic potential in rats. These data provide important information about the drug safety during pregnancy.

Safety evaluation of ondansetron after gestational exposure on male reproductive parameters in rats.

Ondansetron is a 5HT3 receptor antagonist widely used to treat hyperemesis gravidarum, although its safety is still questionable. Since 5HT3 receptors, which are the target of this drug, can interfere with brain development through changes in neurotransmitter levels, this study evaluated whether the prenatal exposure to this drug could compromise reproductive and behavioral parameters in male offspring. Pregnant rats were treated with ondansetron (1.7 and 2.5 mg/kg/body weight; gavage), from gestational day 1-21. No exposure-related changes in clinical signs, body weight, food consumption, pregnancy length, and necropsy findings were observed in dams. Ondansetron exposure did not alter the anogenital distance or age of preputial separation in male offspring. Similarly, males exposed to therapeutic doses of ondansetron did not exhibit changes in play behavior. In adulthood, there were no changes in sperm parameters, as well as in testosterone level, sexual behavior and fertility. Furthermore, ondansetron did not interfere with testicular and epididymal histology, and with androgen receptor expression in hypothalamus. In conclusion, prenatal exposure to ondansetron did not cause maternal toxicity, as well as did not interfere with reproductive parameters of male offspring, indicating its safety after gestational exposure in rats.

Pregnancy outcome following in-utero exposure to ondansetron: A prospective comparative observational study.

The objective of the study was to evaluate the rate of major congenital anomalies after first trimester exposure to ondansetron for nausea and vomiting of pregnancy (NVP). The design is a prospective, comparative, observational cohort study, performed at the Israeli Teratology Information Service between 2010 and 2014. Follow-up was obtained for 195 ondansetron-exposed, 110 metoclopramide-exposed, and 778 pregnancies with non-teratogenic exposure (NTE). The overall rate of major anomalies did not significantly differ between the groups [4/200 = 2.0 % (ondansetron), 1/109 = 0.9 % (metoclopramide), and 13/731 = 1.8 % (NTE)]. All the anomalies in both the ondansetron and metoclopramide groups, and 6/13 anomalies in the NTE group, were cardiac septal defects most of which spontaneously resolved. Both ondansetron (adjHR = 0.29, 95 % CI 0.10-0.80) and metoclopramide (adjHR = 0.27, 95 % CI 0.08-0.86) were associated with lower miscarriage rate compared to NTE. Based on the present study, ondansetron during pregnancy is not associated with an increased risk for overall major anomalies, nor for clinically important cardiac defects. It may be a reasonable alternative for women with severe NVP who do not respond to first line medications.

Major Congenital Malformation Risk After First Trimester Gestational Exposure to Oral or Intravenous Ondansetron.

Ondansetron is a 5-HT3 receptor antagonist that has been approved for the prevention of nausea and vomiting associated with cancer chemotherapy, radiotherapy, and surgery. Ondansetron has also been studied in the treatment of many neuropsychiatric and medical conditions. The drug is commonly used off-label to treat nausea and vomiting of pregnancy (NVP) and hyperemesis gravidarum (HG). Ondansetron crosses the placental barrier, and concerns have been expressed that using ondansetron for NVP/HG during the first trimester of pregnancy may increase the risk of major congenital malformations (MCMs) in the offspring. In this context, findings from a meta-analysis of 6 cohort and 2 case-control studies, read along with the results of subsequently published cohort (n = 3) and case-control (n = 1) studies, suggest that a signal does exist to associate early gestational exposure to ondansetron with an increased risk of heart defects and orofacial defects. Arguments both for and against confounding by indication have been proposed to explain these findings. Nevertheless, even if ondansetron is causally implicated in MCM risk, the absolute increase in risk, such as for orofacial clefts (by 0.03%) and ventricular septal defect (by 0.3%), is small. These small risks should be balanced against the risks associated with inadequately treated NVP/HG, and decision-making must be shared between clinician and patient. Repeated fetal scanning during the second trimester can help in the early detection of malformations, if present.

Ondansetron and teratogenicity in rats: Evidence for a mechanism mediated via embryonic hERG blockade.

The potent hERG channel blocking drug ondansetron is used off-label for treatment of nausea and vomiting in early pregnancy. Some human epidemiological studies have associated ondansetron with fetal cardiovascular defects and orofacial clefts. This study investigated the effects of ondanestron on embryonic heart rhythm of gestational day (GD) 13 rat embryos in vitro and then integrated the results with published animal teratology, and animal and human pharmacokinetic studies to perform a risk evaluation. Ondansetron caused concentration dependent bradycardia and arrhythmia. Cardiovascular malformations in rats occurred at exposures slightly higher than those in early human pregnancy. Together the results suggest that ondansetron can have teratogenic potential in rats and humans mediated via hERG block and severe heart rhythm disturbances in the embryo. The risk may be increased in human pregnancy if additional risk factors are present such as hypokalemia.

Additive Proarrhythmic Effect of Combined Treatment with QT-Prolonging Agents.

Drug combinations may elevate the risk of proarrhythmia. The aim of the present study was to investigate whether combinations of non-cardiovascular agents induce an additive increase in the proarrhythmic risk. In 12 female rabbit hearts, a drug combination of cotrimoxazole (300 µM), ondansetron (5 µM) and domperidone (1 µM) was infused after obtaining baseline data. In another 13 hearts, a combination of cotrimoxazole (300 µM), ondansetron (5 µM) and erythromycin (300 µM) was infused. Monophasic action potentials and ECG displayed a significant QT prolongation in all groups. This was accompanied by a significant increase in action potential duration. Of note, addition of each drug resulted in a further increase in the QT interval. Furthermore, a significant elevation of spatial dispersion of repolarization was observed. Lowering of potassium concentration in bradycardic AV-blocked hearts provoked early afterdepolarizations and torsade de pointes (TDP) in both study groups. Under baseline conditions, no episodes of TDP recorded. After administration of the first agent, TDP occurred in 5 of 12 hearts (37 episodes) and 5 of 13 hearts (26 episodes), respectively. After additional infusion of the second drug, TDP were recorded in 7 of 12 hearts (55 episodes) and 8 of 13 hearts (111 episodes). After additional infusion of the third drug, TDP occurred in 11 of 12 hearts (118 episodes) and 9 of 13 hearts (88 episodes). Combined treatment with several non-cardiovascular QT-prolonging agents resulted in a remarkable occurrence of proarrhythmia. An additive and significant prolongation of cardiac repolarization combined with an increased spatial dispersion of repolarization represents the underlying electrophysiological mechanism.

Severe Proarrhythmic Potential of the Antiemetic Agents Ondansetron and Domperidone.

The potential of ondansetron and domperidone, both clinically established antiemetic agents, to increase the QT-interval has been described in several case reports. Therefore, the aim of the present study was to investigate whether these drugs may provoke polymorphic ventricular tachycardia in a sensitive experimental model of drug-induced proarrhythmia. In 10 female rabbits, ondansetron (1, 5 and 10 µM, n = 10) or domperidone (0.5, 1 and 2 µM, n = 8) was infused after obtaining baseline data. Eight endo- and epicardial monophasic action potentials and a simultaneously recorded 12-lead ECG reproduced the clinically observed QT-prolongation (ondansetron: 1 µM:+17 ms, 5 µM:+41 ms, 10 µM:+78 ms, p < 0.01; domperidone: 0.5 µM:+57 ms, 1 µM:+79 ms, 2 µM:+99 ms, p < 0.01). This was accompanied by a significant increase in action potential duration at 90% of repolarization. Administration of both agents also increased dispersion of repolarization (ondansetron: 1 µM:+12 ms, 5 µM:+17 ms; 10 µM:+18 ms, p < 0.05; domperidone: 0.5 µM:+19 ms, 1 µM:+27 ms; 2 µM:+23 ms p < 0.05). Lowering of potassium concentration in bradycardic AV-blocked hearts provoked early afterdepolarizations (EADs) in 9 of 10 ondansetron-treated hearts and induced polymorphic ventricular tachycardia (VT) resembling torsade de pointes in 7 of 10 ondansetron-treated hearts (86 episodes). Under the influence of domperidone, EAD and polymorphic VT occurred in 7 of 8 hearts (131 episodes). In the present study, both ondansetron and domperidone demonstrated a severe proarrhythmic potential. A significant prolongation of cardiac repolarization as well as a marked increase in spatial dispersion of repolarization represents the underlying electrophysiologic mechanisms. These results imply that application of ondansetron should be handled carefully. For regular administration, ECG monitoring should be mandatory.

Publishing negative findings and the challenge of avoiding type II errors in studies of suspect teratogens: Example of a recent ondansetron publication.

It is important that negative, as well as positive, studies be published to complete the available picture in areas of scientific inquiry. At the same time, it is critical that the implications of a negative study not be overstated and generalized when major issues of study design and data accuracy may be the reason that no relationship was discovered. The challenge of avoiding type II errors in interpreting negative findings has major public health implications, especially when the relationship of an exposure to birth defects is the concern. This is particularly important when interpreting the report by Fazio et al. (June issue of Reproductive Toxicology) on the relationship of ondansetron exposure to pregnancy outcome and birth defects. This review addresses the study design and conclusions and suggests that an alternative concluding statement would be more apropos, given the limitations of the data.

Ondansetron can enhance cisplatin-induced nephrotoxicity via inhibition of multiple toxin and extrusion proteins (MATEs).

The nephrotoxicity limits the clinical application of cisplatin. Human organic cation transporter 2 (OCT2) and multidrug and toxin extrusion proteins (MATEs) work in concert in the elimination of cationic drugs such as cisplatin from the kidney. We hypothesized that co-administration of ondansetron would have an effect on cisplatin nephrotoxicity by altering the function of cisplatin transporters. The inhibitory potencies of ondansetron on metformin accumulation mediated by OCT2 and MATEs were determined in the stable HEK-293 cells expressing these transporters. The effects of ondansetron on drug disposition in vivo were examined by conducting the pharmacokinetics of metformin, a classical substrate for OCTs and MATEs, in wild-type and Mate1-/- mice. The nephrotoxicity was assessed in the wild-type and Mate1-/- mice received cisplatin with and without ondansetron. Both MATEs, including human MATE1, human MATE2-K, and mouse Mate1, and OCT2 (human and mouse) were subject to ondansetron inhibition, with much greater potencies by ondansetron on MATEs. Ondansetron significantly increased tissue accumulation and pharmacokinetic exposure of metformin in wild-type but not in Mate1-/- mice. Moreover, ondansetron treatment significantly enhanced renal accumulation of cisplatin and cisplatin-induced nephrotoxicity which were indicated by increased levels of biochemical and molecular biomarkers and more severe pathohistological changes in mice. Similar increases in nephrotoxicity were caused by genetic deficiency of MATE function in mice. Therefore, the potent inhibition of MATEs by ondansetron enhances the nephrotoxicity associated with cisplatin treatment in mice. Potential nephrotoxic effects of combining the chemotherapeutic cisplatin and the antiemetic 5-hydroxytryptamine-3 (5-HT3) receptor antagonists, such as ondansetron, should be investigated in patients.

Seasonal modulation of the 8-and 24-hour rhythms of ondansetron tolerance in mice.

Ondansetron (Zophren((R))) is a serotonin 5HT(3)-receptor antagonist used primarily to control nausea and vomiting caused by cytotoxic chemo-and radio-therapy. Tolerance to this drug shows both 24 and 8 h periodicities. In this framework, this study aimed to determine whether these ondansetron tolerance rhythms are modulated by season. The chronotoxic effect of a fixed dose (3.5 mg/kg, i.p.) of the drug was investigated with reference to both time of the day and year dependencies. Season-related studies were performed on 560 male Swiss mice, 10 to 12 wks old, synchronized with L:D=12:12 for three weeks. During a 1 yr span (2005), four 24 h studies were performed with a single dosing time at 1, 7, 13, and 19 hours after light onset (HALO), respectively. Tolerance was assessed daily during a 40-day span after acute ondansetron treatment. Both chi(2) test and cosinor methods were used to analyze the time series data. Statistically significant dosing time-dependent changes were validated in both yearly and daily time scales. The 24 h mean survival rate peaked in spring (92%) compared to fall (72%), the 20% difference being statistically significant (chi(2) test with p<0.05 and cosinor with p<0.0001 for seasonal rhythm detection and with a peak time, Ø,=April 3+/-6.6 days). A 24 h rhythm was also detected in each of the seasonal time points. However, the curve pattern was monophasic in fall as well as spring. In fall, a large amplitude (A) circadian rhythm was detected that peaked at 19 HALO, while in the spring, a small circadian rhythm was detected that peaked at 1 HALO. The curve pattern was biphasic in summer (with large A) and in winter (with a small A). The existence of two peaks of equal magnitude in winter (100% survival rate) and in summer (100% and 90%) suggests the presence of both circadian and ultradian rhythms rather than an ultradian component of the 24 h period. The seasonal modulation of ondansetron circadian chronotolerance seems to involve several rhythm parameters: season-related changes in the 24 h mean (M), amplitude (A), acrophase location (Ø), as well as bimodal curve patterns including the coexistence of rhythms with respectively 24 and 8 h periods in winter and summer. In conclusion, tolerance to ondansetron varies not only according to the 24 and 8 h periods but also according to seasons, which suggests the complexity of ondansetron toxicity rhythms. Seasonal modulation of ondansetron tolerance may also influence the strategies of chemo-and chrono-therapy, and it is therefore necessary to take it into account in clinical drug-delivery protocols to minimize side effects of cytotoxic anticancer and antiemetic agents.

Circadian rhythms in toxic effects of the serotonin antagonist ondansetron in mice.

The aim of the study was to learn whether the lethal and the motor incoordination (ataxia) side effect of ondansetron (Zophren) administration is dosing-time dependent. Ondansetron is a serotonin 5-HT3 receptor antagonist used primarily to control nausea and vomiting arising from cytotoxic chemo- and radiotherapy. A total of 210 male Swiss mice 10 to 12 weeks of age were synchronized for 3 weeks by 12 h light (rest span)/12 h dark (activity span). Different doses of ondansetron were injected intraperitoneally (i.p.) at fixed times during the day to determine both the sublethal (TD50) and lethal (LD50) doses, which were, respectively, 3.7 +/- 0.6 mg/kg and 4.6 +/- 0.5 mg/kg. In the chronotoxicologic study a single dose of ondansetron (3.5 mg/kg, i.p.) was administered to different and comparable groups of animals at four different circadian stages [1, 7, 13, and 19 h after light onset (HALO)]. The lethal toxicity was statistically significantly dosing time-dependent (chi2 = 21.51, p < 0.0001). Drug dosing at 1 HALO resulted in 100% survival rate whereas drug dosing at 19 HALO was only one-half that (52%). Similarly, lowest and highest ataxia occurred when ondansetron was injected at 1 and 19 HALO, respectively (chi2 = 22.24, p < 0.0001). Effects on rectal temperature were also dosing-time related (Cosinor analysis, p < 0.0001). The characteristics of the waveform describing the temporal patterns differed between the studied variables, e.g., lethal toxicity and survival rate showing two peaks and rectal temperature showing one peak in the 24 h time series waveform pattern. Cosinor analysis also revealed a statistically significant ultradian (tau = 8 h) rhythmic component in the considered variables. Differences in curve patterns in toxicity elicited by ondansetron on a per end point basis are hypothesized to represent the phase relations between the identified 24 h and 8 h periodicities.

Double-blind, randomized, parallel-group study on the efficacy and safety of oral granisetron and oral ondansetron in the prophylaxis of nausea and vomiting in patients receiving hyperfractionated total body irradiation.

The efficacy and safety of granisetron and ondansetron for the prophylaxis of nausea and vomiting resulting from hyperfractionated total body irradiation (TBI) were assessed. Thirty-four patients randomly received double-blind, oral granisetron (2 mg, 1 h before first daily fraction of radiation) or ondansetron (8 mg, 1.5 h prior to each fraction of TBI). Ninety patients who received the same TBI regimen prior to bone marrow transplantation (BMT), but no 5-HT3-receptor antagonist, were identified and comprised the historical control group. By design, this study was only powered to show a difference between each of the active treatment groups and the historical control group. Significantly more patients given granisetron (33.3%) or ondansetron (26.7%) had zero emetic episodes over 4 days, the primary efficacy end point, than those in the historical control group (0%) (P < 0.01; intent-to-treat). Secondary efficacy end points were also evaluated. During the first 24 h, significantly more patients taking granisetron (61.1%) or ondansetron (46.7%) had zero emetic episodes than patients in the historical control group (6.7%) (P < 0.01). Complete emetic control (no emesis or rescue antiemetic) over 4 days was more frequent in patients taking granisetron (27.8%) or ondansetron (26.7%) compared with the historical control group (0%) (P < 0.01). Significantly fewer patients taking granisetron (18/18), but not those taking ondansetron (12/15), experienced more than five emetic episodes during the 4 days of the study compared with the historical control group (40/90; P < 0.01). Oral granisetron and ondansetron are safe and effective for the prevention of nausea and vomiting resulting from TBI.

Ondansetron vs metoclopramida para la profilaxis de náusea y vómito postoperatorio después de cirugía abdominal / Ondansetron versus metoclopramida for the náusea prophylaxis and postoperating vomit after abdominal surgery

Una gran incidencia de náusea y vómito postoperatorio (NVPO) es reportada en pacientes programados pra cirugía abdominal. Este estudio fue realizado para comparar la eficacia de ondansetrón versus metoclopramida en la prevención de NVPO en pacientes sometidos a anestesia general para cirugía abdominal. Se realizó un estudio clínico, experimental, randomizado, simple ciego, en 60 pacientes de ambos sexos divididos en dos grupos de 30 cada uno, entre 15 y 64 años, clasificados con estado físico ASA I, II, III, que recibieron en forma aleatoria 4mg IV de ondansetrón (grupo A) y 10 mg IV de metoclopramida (grupo B) cinco minutos antes de la finalización de la cirugía. La misma técnica de anestesia general fue utilizada la cual consistió en fentanylo, tiopental, vecuronio, ethrane y oxígeno. Se revirtió el bloqueo neuromuscular con prostigmine y atropina. La náusea y el vómito así como los efectos adversos de éstas drogas, tensión arterial, frecuencia cardíaca se valoraron continuamente durante las primeras seis horas después de la cirugía. Se estableció además la relación entre sexo, tipo de cirugía (electiva o de urgencia), procedimiento (abierto o laparoscópico), diagnóstico (patología gastrointestinal o ginecológica), tiempo anestésico y náusea y vómito. La náusea fue valorada por una escala objetiva de 0 a 3, y el vómito por el número de presentación. El análisis estadístico fue realizado usando test de chi cuadrado para las variables cualitativas y test de comparación proporcional, de medias y de varianza para las variables cuantitativas. La incidencia de NVPO fue igual con ondansetrón y con metoclopramida, sin existir diferencia significativa (pNS). Se registró una disminución de la frecuencia cardíaca mayor con metoclopramida que con ondansetrón a las cuatro horas posteriores a la cirugía (p 0.005). Las mujeres presentaron una tendencia mayor a la náusea que los varones (p 0.19). Para los otros parámetros no se estableció ninguna diferencia estadísticamente significativa. Concluimos en este estudio que ondansetrón y metoclopramida son igualmente efectivos como profilácticos en el control de NVPO durante las primeras seis horas de postoperatorio.

Ondansetron modulates GABA(A) current of rat central nervous system neurons.

We examined the effect of ondansetron, a 5-HT3 receptor antagonist, on the whole cell current response of freshly isolated hypothalamic and hippocampal neurons of rats to gamma-aminobutyric acid (GABA). The nystatin perforated patch technique was used to minimize run-down of the GABA current. While 1-150 microM ondansetron had no effect on membrane conductance, co-application with agonist reversibly depressed the maximal end GABA current. The concentration-response relation of GABA reveals a non-competitive mechanism. However, the inhibitory effect was more potent when ondansetron was co-applied with lower concentrations of GABA: i.e., the ondansetron concentration needed to depress the current induced by 5 microM GABA to half amplitude was 7 microM compared to 28 microM for the current induced by 10 microM GABA. Analysis of the current-voltage relationship with and without ondansetron indicated that the effect of ondansetron is not voltage dependent. Current-voltage relations also showed that the effect of ondansetron was not due to activation of a GABA-independent current because the reversal potentials were the same with and without ondansetron. The present data suggest that ondansetron's suppression of GABA-activated current may be the molecular basis of ondansetron-induced seizures observed in vivo.

Gastric motility and mucosal ulcerogenic responses induced by prokinetic drugs in rats under prostaglandin-deficient conditions.

The present study was performed to examine whether gastric prokinetic drugs may induce damage in the rat stomach under normal and prostaglandin (PG)-deficient conditions. Male SD rats fasted for 18 hr were administered subcutaneously with three different prokinetic drugs such as metoclopramide (3-60 mg/kg), ondansetron (0.3-3 mg/kg), and cisapride (3-30 mg/kg). Half the number of these animals were pretreated with indomethacin (5 mg/kg) subcutaneously for induction of PG deficiency in the stomach. Administration of these drugs increased gastric motor activity in a dose-dependent manner and expedited gastric emptying at lower doses than those affecting gastric motility; the potency of the hypermotility effect was in the following order: metoclopramide = ondansetron > cisapride. None of these drugs alone caused gross damage in the stomach, although whitish rough areas were observed in the gastric mucosa along the folds. In the rats pretreated with indomethacin, however, both metoclopramide and ondansetron provoked multiple hemorrhagic lesions in the gastric mucosa. Indomethacin at this dose showed over 90% inhibition of cyclooxygenase activity without causing any damage in the stomach, and this PG-deficient effect was not affected by coadministration with the prokinetic drugs. The mucosal ulcerogenic responses induced by metoclopramide in the presence of indomethacin were significantly inhibited by prior administration of atropine (1 mg/kg) or PGE2 (300 micrograms/kg) at doses that inhibited gastric hypermotility induced by metoclopramide. These results suggest that: (1) gastric prokinetic drugs induce damage in rat stomachs under PG-deficient conditions at the doses that enhance gastric motility and emptying but not at the doses that expedite gastric emptying only, and (2) gastric hypermotility has the potential to cause gross damage in the stomach, supporting the importance of gastric motility as a pathogenic element of gastric lesions.

GYKI-46 903, a non-competitive antagonist for 5-HT3 receptors.

The effects of GYKI-46 903 ((+)endo-4-propionyloxy-6-(4-fluorophenyl)-1-azabicyclo [3.3.1]non-6-ene HCl), on 5-HT3 receptors have been studied and compared with ondansetron in peripheral organs in vitro and in vivo, and in a receptor binding assay in membranes prepared from rat cerebral cortex. GYKI-46 903 was found to be a non-competitive antagonist at 5-HT3 receptors present in non-stimulated longitudinal muscle strip of guinea-pig ileum (pD2' against serotonin = 5.54), and also in 5-methoxytryptamine-pretreated electrically stimulated ileal preparations (pD2' against serotonin = 5.26). On the contrary, ondansetron was found to be a competitive antagonist for 5-HT3 receptors; the pA2 value against serotonin was 7.40 in non-stimulated ileum, and it was 7.08 in electrically stimulated ileal preparation pretreated with 5-methoxytryptamine. In displacement studies, the pIC50 values of GYKI-46 903 and ondansetron against [3H]granisetron binding to rat cerebral cortex membranes were 6.91 and 8.58 respectively. GYKI-46 903, when administered by intravenous infusion, antagonized the decrease in heart rate evoked by serotonin (Bezold-Jarisch reflex) in anaesthetized rats, and the maximal reversal was less than 50%. This was in striking contrast with ondansetron, which, after intravenous injection, completely antagonized the serotonin-induced bradycardia with an ID50 value of 3.28 ug/kg. These data classify GYKI-46 903 as a non-competitive antagonist for 5-HT3 receptors.

Development of an in vitro hepatotoxicity assay for assessing the effects of chronic drug exposure.

We have used the human hepatoma cell line HepG2 to compare the hepatotoxic effects of acute (24 hr) and chronic (up to 10 days) exposure to amitriptyline, paracetamol and ondansetron. In acute exposure studies, hepatotoxicity was assessed by the sulforhodamine B protein staining method, where amitriptyline and paracetamol produced 50% hepatotoxicity at concentration of 30 microM and 7 mM, respectively, while ondansetron was non-hepatotoxic at 100 microM, the highest concentration used. In chronic exposure studies, the morphology of HepG2 cells was assessed by phase microscopy every 2 days and the compounds, at concentrations determined from the acute assay, were added fresh every 2 days. Chronic exposure to amitriptyline and paracetamol produced significant morphological changes in HepG2 cells at 3 microM and 1 mM respectively, concentrations which had no significant effect in the acute assay. Ondansetron (100 microM) produced only slight morphological changes in the cells after 10 days of culture. The combination of acute and chronic drug exposure assays with HepG2 cells represents novel in vitro systems for the hepatotoxicological assessment of drugs intended for human use.