Antimicrobial-induced torsades de pointes: towards risk stratification by data mining of the FDA Adverse Event Reporting System.

Introduction: In order to critically evaluate the risk of antimicrobial-induced torsades de pointes (TdP) in clinical practice, we investigated spontaneous reports submitted to the Adverse Events Reporting System (AERS). Methods: From the public version of the AERS reports of TdP from January 2004 through December 2007 were retrieved. Suspected and interacting drugs were identified. The absolute number of cases, disproportionality in reporting rate and published evidence were evaluated for each antimicrobial drug. Results: Over the –year period, reports of TdP were 275 involving 20 antibacterials, 8 antifungals and 18 antivirals. Antimicrobials more frequently reported were levofloxacin (49) and moxifloxacin (26) among antibacterials, fluconazole (41) and voriconazole (13) among antifungals, lamivudine (7) and nelfinavir (6) among antivirals. A significant disproportionality was observed for 16 compounds including several macrolides, fluoroquinolones, triazole antifungals, antiretrovirals, linezolid, cotrimoxazole and caspofungin. Since the amount of evidence, disproportionality signals for macrolides, fluoroquinolones and azoles antifungals should be viewed as ‘expected’. By contrast, signals generated by linezolid, cotrimoxazole and caspofungin were considered ‘unexpected’. These drugs may belong to a hypothetical ‘black list’. Compounds with no published evidence, lack of signal and extensive use (e.g. amoxicillin) may be included in a provisional ‘white list’. A ‘grey list’ could include all drugs with questionable data on TdP liability. Conclusions: The AERS database represents an important source to detect ‘unexpected’ signals of TdP. Although further investigations are warranted, this provisional risk stratification represents a step towards an appropriate clinical use of antimicrobials, especially in patients susceptible to arrhythmia.