#NephJC Chat
Tuesday October 22nd, 2024 at 9 pm Eastern (AEST = October 23rd, 12 am)
Wednesday October, 23th, 2024, at 9 pm by Indian Standard Time/ 3:30 pm GMT (AEST = October 24th, 2:30 am)
Journal of the American Society of Nephrology 35(6): p 761-771, June 2024. |
DOI: 10.1681/ASN.0000000000000336
Ondansetron and the Risk of Sudden Cardiac Death among Individuals Receiving Maintenance Hemodialysis
Sherin Ismail, Michele Jonsson Funk, Jennifer E Flythe
PMID: 38409683
Introduction
Retch, regurgitate, throw up, heave, puke, barf, toss one’s cookies; just the mere mention of words associated with nausea and vomiting, or witnessing an event, is enough to churn some stomachs. Nausea is an extremely common problem for patients on dialysis and may be related to uremia, hypotension, dialysis disequilibrium syndrome, diabetes with gastroparesis, anxiety, or even diet and medications. Since most nephrologists will encounter a patient with nausea almost daily, the question remains: which antiemetic is safest for patients on maintenance hemodialysis? Unfortunately, many treatments for nausea can pose serious risks, including QT prolongation, Torsades de Pointes, and sudden cardiac death. Patients on hemodialysis are particularly vulnerable, with sudden cardiac death being the leading cause of mortality in this group, driven by factors such as cardiovascular disease, frequent electrolyte shifts, and polypharmacy (Assimon et al., JAHA 2020).
Despite these risks, no studies have compared the safety of various antiemetics in patients on dialysis. Unfortunately, QT-prolonging antiemetics are frequently prescribed to this group, sometimes by non-nephrology clinicians, increasing the risk of cardiac events (Wang et al., JAMA 2024). Ondansetron, a widely used antiemetic, is associated with dose-dependent QT prolongation and Torsades de Pointes, leading the FDA in 2012 to limit its single intravenous dose to 16 mg. Ondansetron was originally developed to manage nausea and vomiting associated with chemotherapy, radiotherapy, and postoperative nausea and vomiting, but nowadays it is frequently used (and considered safe) for other causes of nausea as well. Other commonly used antiemetics include metoclopramide, promethazine, and prochlorperazine which have a lesser QT-prolonging potential compared with ondansetron, but carry their own risks and side effect profiles.
However, QT prolongation might be only considered a surrogate marker, and the real test is whether these medications are associated with worse clinical outcomes. In this case, the outcome of concern is sudden cardiac death, which is not infrequent for patients on dialysis. Data from the Dialysis Outcomes and Practice Patterns (DOPPS) registry showed that sudden death is common among hemodialysis patients, around 20-30% with some variation: United States (33% of all deaths), Japan (23%), Australia/New Zealand (19%), and Canada (18%) (Jadoul et al, CJASN 2012). To answer the question of clinical outcomes with these drugs, let’s discuss a retrospective cohort study comparing the association of oral ondansetron to other antiemetics (promethazine, prochlorperazine, metoclopramide) and the risk of sudden cardiac death in patients on hemodialysis.
The Study
Methods
This is an observational, cohort study examining the association between the initiation of ondansetron versus the initiation of a comparator antiemetic (promethazine, metoclopramide, prochlorperazine) and the 10-day risk of sudden cardiac death among individuals receiving maintenance in-center hemodialysis.
Data were obtained from the United States Renal Data System (USRDS), a national surveillance system that collects, analyzes, and distributes information on individuals with kidney failure in the United States (NephJC discussion of the 2023 report).
The following figures show how the patients with new use of antiemetics were selected. They identified patients who met the selection criteria by initiation of ondansetron, or a comparator antiemetic, after a 30-day washout period of any of the study drugs. The index date was defined as the date of study antiemetic initiation. Follow-up began on the index date. Baseline covariates were identified in a 180-day period before index date.
Figure 1. Study design, from Ismail et. al, JASN, 2024
Study Antiemetic Medications
New users of antiemetics ondansetron, promethazine, metoclopramide, and prochlorperazine were identified. CredibleMeds and the FDA’s Adverse event reporting system have classified these medications according to their risk of developing Torsades de Pointes.
Based on this information, “new use” of ondansetron was compared with “new use” of antiemetic drugs with lesser Torsades de Pointes risk. Oral antiemetics were studied because oral formulations are prescribed more frequently than intravenous formulations in the outpatient setting.
Study Outcomes
Primary Outcome
Ten-day risk of sudden cardiac death defined using the USRDS definition of death due to cardiac arrhythmia or cardiac arrest.
Secondary Outcomes
A composite of sudden cardiac death or hospitalized ventricular arrhythmia.
Cardiovascular mortality.
In a sensitivity analysis, hospitalization for fracture as a negative control outcome was studied.
Statistical Analysis
The association between the initiation of ondansetron, versus other antiemetics, and the 10-day risk of sudden cardiac death was evaluated. The authors selected 10 days as the duration of the primary follow-up because the median supply of the studied antiemetics ranged from 7-30 days. Sensitivity analysis was done for a 30-day follow-up. Intention to treat analysis was used, similar to a randomized controlled trial. The patients were followed from the index date to the first occurrence of the study outcome or a censoring competing event (kidney transplantation, loss of Medicare coverage, change in hemodialysis modality, completion of 10-day follow-up, study end).
The authors used inverse probability of treatment (IPT) weighting to control for confounders. In addition, multivariable logistic regression models were used to estimate propensity scores (the predicted probability of initiation ondansetron versus a comparator antiemetic conditional to baseline characteristics). With the propensity scores, they calculated IPT weights, to balance baseline covariates between the two groups. In the unweighted and weighted populations, they used Fine and Gray models to estimate subdistribution hazards ratios (HRs) and 95% confidence intervals (CIs) treating non-sudden cardiac death as a competing event.
Subgroup analyses were conducted in the following: a) age (<65 years and >65 years), b) sex, c) baseline heart failure, d) baseline arrhythmic condition, e) baseline hepatic disease, f) baseline use of at least one medication with known risk of Torsades de Pointes, g) baseline use of at least one medication with possible risk of Torsades de Pointes. Propensity scores and IPT weights were also used to control potential confounders.
The following sensitivity analyses were performed: 30 30-day follow-up period, longer washout periods for the study antiemetics (60, 90, 180 days), and a negative control outcome analysis in which they evaluated the 10-day risk of hospitalized fracture.
Funding
J.E. Flythe: National Heart, Lung, and Blood Institute.
Results
Study Population
A total of 574,511 patients on dialysis who were newly started on antiemetic between 2007 and 2019, as per USRDS, were assessed for eligibility. Of these, 119,254 patients receiving in-center hemodialysis and also had new use of a study antiemetic between 2012 and 2019 were included. They were assigned to the ondansetron group (n=64,978) or the comparator antiemetic group (n=54,276). Patients who were on dialysis for 90 days or less (n=16,621) and those who received non-oral formulation of study antiemetic (n=7,252) were excluded among other criteria. The study also excluded patients who received index study antiemetic prescriptions before 2012 (n=90,400) as the FDA label warning on ondansetron predisposing to Torsades de Pointes was updated in 2012.
Supplemental Figure 1. Flow diagram depicting study Cohort assembly, from Ismail et. al, JASN 2024.
The study patients had a mean age of 60 ± 15 years, with 55% females. Diabetes mellitus was the most common primary cause of kidney failure (52%), followed by hypertension (26%). Comorbid cardiac conditions were common and included 45% of patients with ischemic heart disease, 44% with heart failure, 27% with arrhythmia, and 10% with conduction disorder. The median time on hemodialysis was 2.2 years. Approximately 20% of patients were using one or more medications with a known risk of Torsades de Pointes (according to The CredibleMeds website classification).
Table 1. Select baseline characteristics of ondansetron and comparator antiemetic new users, from Ismail et. al, JASN 2024.
Compared with the comparator antiemetic group, individuals in the unweighted cohort ondansetron group were older (mean age 62 years versus 58 years), had a higher prevalence of baseline arrhythmia (29% versus 25%), were less likely to have diabetes mellitus (72% versus 75%), and more likely to be using one or more medications with a possible risk of Torsades de Pointes (15% versus 6%).
All the baseline characteristics were effectively similar between the two groups after the inverse probability of treatment weighting.
Supplemental Figure 2. Propensity score distributions, from Ismail et al, JASN 2024.
The median pill strength and days of supply for ondansetron and comparator antiemetics were as follows:
Primary Outcome
Out of the total 198 sudden cardiac deaths that occurred within 10 days of the follow-up period, 133 deaths occurred in the ondansetron group and 65 deaths occurred in the comparator antiemetic group. The incidence of sudden cardiac death was 2.1 events per 10,000 person-days in new ondansetron users, compared to 1.2 events per 10,000 person-days in new users of comparator antiemetics.
Table 2. Association between initiation of ondansetron versus a comparator antiemetic and 10-day outcomes. From Ismail et. al, JASN 2024.
There were higher absolute and relative 10-day risks of sudden cardiac death with ondansetron initiation compared to comparator antiemetic initiation (adjusted HR, 1.44 [95% CI, 1.08 to 1.93]; adjusted risk difference 0.06% [95% CI, 0.01% to 0.11%]).
Figure 2. Ten-day risk of sudden cardiac death among patients initiating ondansetron versus comparator antiemetic. (Ismail et. al, JASN 2024) CI, confidence interval; HR, hazard ratio; RD, risk difference.
Secondary Outcomes
Compared with new users of comparator antiemetics, ondansetron new users had a higher 10-day risk of cardiovascular mortality (adjusted HR, 1.37 [95% CI, 1.05 to 1.77]), and the composite outcome of sudden cardiac death or ventricular arrhythmia (adjusted HR, 1.36 [95% CI, 1.04 to 1.78]). The incidence rate of the composite outcome of sudden cardiac death or ventricular arrhythmia was 2.2 events per 10,000 person-days among ondansetron new users compared with 1.3 events per 10,000 person-days among comparator antiemetic new users. The incidence rate of cardiovascular mortality was 2.4 events per 10,000 person-days among ondansetron new users compared with 1.5 events per 10,000 person-days among comparator antiemetic new users.
Subgroup Analyses
Clinically relevant subgroups for primary outcome included a history of heart failure, baseline arrhythmias, use of medications with the risk of Torsades de Pointes, and others. Although the analysis suggests a higher risk of the primary outcome in the ondansetron group compared to other antiemetic groups across most subgroups, subgroup analysis can be underpowered. The limited number of events (198 total SCD events out of 120,000 patients or 0.16%) makes these results potentially less reliable.
Figure 3. Association between the initiation of ondansetron versus comparator antiemetic and the 10-day risk of sudden cardiac death within clinically relevant subgroups. From Ismail et. al, JASN 2024.
Sensitivity Analyses
Study outcomes at the 30-day follow-up period were consistent with the primary results. The analysis of hospitalized fractures as a negative control outcome, in a cohort excluding patients with a history of cancer or fractures, showed that ondansetron initiation (compared to other antiemetics) had an aHR of 1.17 (95% CI, 0.95–1.43) and an aRD of –0.06% (95% CI, –0.02% to 0.14%), thereby proving no unexpected association. Analyses of the risk of sudden cardiac death over progressively later time intervals showed that the association between ondansetron initiation (versus other antiemetics) and the risk of sudden cardiac death weakened over time, approaching no significant difference.
Supplemental Table 9. Sensitivity analyses. From Ismail et al, JASN 2024.
Discussion
Ondansetron is an antiemetic that antagonizes serotonin at 5-hydroxytryptamine receptors (5-HT3). It is completely and rapidly absorbed after oral administration and it’s first detected in plasma after 30 minutes of a 8mg oral dose. Ondansetron is extensively metabolized by the liver and this accounts for more than 95% of its clearance from the body. Renal clearance accounts for less than 5%, which may give a false sense of security and safety in patients with renal impairment. However, no pharmacokinetic studies have specifically been carried out in the ESKD population to date.
In 2011 the FDA issued a communication warning that ondansetron may induce fatal arrhythmias, and in 2012 they issued an update linking the risk of QT prolongation to the administration of a 32 mg intravenous dose. With this in mind, recommendations have urged to avoid ondansetron in patients with congenital long QT syndrome. ECG monitoring and serum metabolic panel screening are recommended to be performed in all potentially susceptible patients with known electrolyte abnormalities, congestive heart failure, bradyarrhythmias, or patients taking other drugs or products that lead to QT prolongation. This may sound redundant, but aren't a significant majority of patients on hemodialysis part of this high-risk population? How many of us really follow this recommendation in our routine daily practice?
Ondansetron use in patients on hemodialysis is fairly ubiquitous. Ondansetron has been proven in some small studies (Ljutic et. al, Kidney Blood Press Res. 2002) to be more efficient than metoclopramide to ameliorate nausea and vomiting in patients with uremia. In fact, not only has it been used to treat nausea, but some authors (Ashmore et al. Am J Kidney Dis 2020) have studied ondansetron as a potential therapy for uremic pruritus. Spoiler alert, it was not effective in the treatment of itch associated with ESKD.
Despite the burden of nausea and vomiting in patients in hemodialysis there are no accepted clinical practice guidelines for the treatment of these symptoms. Experts suggest a stepwise approach beginning with non-pharmacological interventions (controlled breathing, acupressure, chewing gum, clear liquids, bland foods, smaller/more frequent meals), and if needed, advancing to drug therapy prescribed at the lowest possible dose. They do not suggest one antiemetic over another, and the selection of antiemetic therapy is often based on the prescriber's preferred drug.
With this in mind, is it safe to use ondansetron in patients on hemodialysis? This study found that patients who initiated ondansetron versus the comparator antiemetics with less QT interval-prolonging potential had a higher 10-day risk of sudden cardiac death (SCD). The findings were similar for additional cardiac outcomes and were consistent across some clinically relevant groups. Among these groups, patients 65 years and older and with heart failure seemed to have higher 10-day relative risks of SCD when taking ondansetron compared to comparator antiemetics, which is consistent with previous findings. (Assimon MM, et al, Circulation, 2020)
Figure 2. Use of ≥1 prescription medication with known TdP risk by hemodialysis patients with and without risk factors for drug‐induced QT prolongation in 2016. From Assimon MM et al, Circulation, 2020.
It should be noted that some groups had discordant results. For example, patients with arrhythmias show an increased, though not statistically significant, risk of SCD when using ondansetron, but this risk was not greater than in those without arrhythmias. Interestingly, patients on other Torsades de Pointes risk medications showed a lower risk of SCD when starting ondansetron compared to the other antiemetics. The authors tried to explain this by implying prescribers were more aware of the potential risk and closely monitored these patients, although this is merely an assumption. A more plausible reason is mentioned by Hung et, al. JASN 2024, explaining how ondansetron is metabolized in the liver by enzymes of the cytochrome P450 (CYP) family and is often prescribed alongside other gastrointestinal drugs that can induce (omeprazole or rabeprazole) or inhibit (cimetidine or ranitidine) CYP enzymes and affect plasma concentrations. Given the frequency of polypharmacy in patients undergoing chronic hemodialysis, it is essential to consider these factors in the analysis. A more comprehensive retrieval of the medication list would contribute to a more accurate estimation of the ondansetron-associated risk in these patients.
It is important to note that prior studies in the general population link the administration of intravenous doses of ondansetron to QT prolongation-induced fatal arrhythmias and evidence of oral ondansetron is less robust. According to the FDA Adverse Events Reporting System and the World Health Organization Individual Safety Case Reports Database, no reports of arrhythmia associated with single oral dose ondansetron have been made. It is also highly likely that non-reported arrhythmias have occurred in people receiving hemodialysis. In this study, median pill strength for ondansetron was 4 mg, and 7 days of supply was the assumed days of treatment. However, there was no specific accounting for the number of doses actually taken. Because ondansetron is a “prn” medication, only taken with the onset of symptoms, it is possible that the patients did not consume any of the medication during the 10 day follow-up period. This is an important confounder that should be taken into consideration.
The authors discussed the plausibility of the findings given the underlying cardiac risk profiles of patients on hemodialysis. These patients have a high prevalence of structural cardiac disease as well as coronary disease and are regularly exposed to shifts in electrolytes relevant to cardiac conduction. The fact that the patients in the study received oral ondansetron, and were still at a higher risk of sudden cardiac death, speaks to the fact that there may be no safe oral or intravenous dose in high risk patients.
Ondansetron and metoclopramide are possibly the most common antiemetics in hemodialysis units. Metoclopramide is used with caution because of the risk of tardive dyskinesia and promethazine, also known as a sedative, has abuse potential. These effects are not anywhere as serious as SCD, which is thankfully rare, but any adverse events can be quite concerning to patients. Although this retrospective cohort study can not definitively associate ondansetron use with SCD, ondansetron use is definitely worth additional caution and consideration in patients on dialysis.
Strengths:
It is a large, national representative cohort.
Use of a new user, active comparator design to minimize confounding and enhance comparability between the study groups.
Sensitivity analysis to assess robustness of results, with the longer follow-up periods results consistent with those of the primary analysis.
Limitations:
The study was observational and even if the propensity score distributions of the comparator antiemetics exhibited overlap, residual confounding may have remained.
The negative control outcome of hospitalized fracture was inconclusive with regard to potential residual confounding.
Study antiemetic initiation was determined using prescription medication claims and this date does not reflect actual antiemetic consumption.
No intravenous dose of ondansetron or the comparator antiemetics was studied, which is plausible in patients on a hemodialysis setting.
It is possible that reverse causation may have influenced findings (an imminent cardiac event could cause nausea/vomiting ). Other possible situations that imply risk for SCD like low EF, insufficient dialysis does and DM complications might have also influenced results.
Conclusion
Although this observational cohort study is not definitive in associating ondansetron with sudden cardiac death, clinicians should be mindful of the QT-prolonging potential in patients on hemodialysis. For patients in which the clinician determines the benefits of using this drug outweighs the potential risks, consideration should be given to performing safety measures like electrolyte and ECG monitoring. Physicians should also review all medications that could have additional potential to cause QT prolongation and fatal arrhythmias.
Summary by
Oscar Durón
Nephrologist, Oviedo, Spain
Nephrology Fellow, Boston, MA
NSMC Interns class of 2024, Pod 1 Filtrate Firebolts