An evaluation of potential signals for ventricular arrhythmia and cardiac arrest with dolasetron, ondansetron, and granisetron in the fda combined spontaneous reporting system/adverse event reporting system.

BACKGROUND: Of the US Food and Drug Administration (FDA)-approved5-hydroxytryptamine type 3 (5-HT3)-receptor antagonists, dolasetron, ondan-setron, granisetron, and palonosetron, only dolasetron and palonosetron have a precaution in their FDA labeling concerning corrected QT interval (QTc) prolongation. At FDA approved doses, QTc prolongation has been observed in clinical trials with some 5-HT3 receptor antagonists (however, palonosetron has been only recently approved, with few published clinical data available). However, due to patient exclusion criteria, such trials with 5-HT3 receptor antagonists may have failed to examine the risk of these agents in "real world" patients with cancer.
OBJECTIVE: The aim of this analysis was to assess the potential risk for selected cardiac adverse events associated with dolasetron, ondansetron, and granisetron use.
METHODS: The FDA combined Spontaneous Reporting System/Adverse Event Reporting System database was analyzed. The process of analyzing such a database for early warnings of potential hazards is known as signal generation. The statistical technique proportional reporting ratio (PRR) was used to aid detection of a potential signal within the database. PRR is the observed proportion of a given adverse event for the drug of interest (the number of events of interest for the drug divided by the total number of reports for the drug) divided by the expected proportion. Through the third quarter of 2002, the database was searched using the preferred term electrocardiogram qt corrected interval prolonged.
RESULTS: One, 3, and 0 cases were reported for dolasetron, ondansetron, andgranisetron, respectively. The number of cases did not satisfy 1 of the 3 criteria we utilized to define a potential signal, the 3 criteria being: 3 or more reported cases of the adverse event, a PRR value of at least 2, and a χ(2) value of >4. As this term may be unlikely to be reported, the database was also searched using the term ventricular arrhythmias and cardiac arrest. The PRR, used as a parameter to detect a potential signal within the database, was 3.23, 1.31, and 1.13 for dolasetron, ondansetron, and granisetron, respectively. The number of observed ventricular arrhythmias and cardiac arrests was ∼3-fold higher with dolasetron compared with the expected value (calculated by dividing the individual agent's total number of events reported by the proportion of adverse events for all agents combined). The results for dolasetron fulfilled the criteria we used to define a potential signal.
CONCLUSIONS: This analysis detected a potential signal for ventricular arrhythmiasand cardiac arrest with dolasetron, but not with ondansetron or granisetron. However, there are limitations of a PRR analysis, which include only measuring cases that have been reported, providing relative frequencies instead of actual rates, and not providing information on the severity of adverse events or causal relationships. In addition, our analysis does not include consideration of concomitant medications, and only 2 search terms were used. Errors in identifying potential signals may also include confounding factors, such as the underlying disease, potential confusion with reporting under trade and generic names, and potential multiple reporting of the same case.