Safety of Voriconazole and Dose Individualization

Sir—In the 15 November issue of Clinical Infectious Diseases, Potoski and Brown [1] suggested there is a need for individual voriconazole dosage modification based on plasma concentrations. Their concerns focus on the possibility of voriconazole-related liver toxicity. More than 2000 patients, most of whom were severely immunocompromised, were treated with voriconazole during the clinical development program, which provided an extensive database for safety analysis. The data on clinically significant abnormal liver function test (LFT) results from 3 pivotal comparative studies [2–4] are presented in table 1. Although the frequency of abnormal LFT results was relatively high, the abnormalities themselves were mostly mild to moderate in severity and led to the discontinuation of voriconazole therapy only rarely (for 3% of patients). Only in patients with esophageal candidiasis was the frequency of abnormal LFT results greater among patients treated with voriconazole than among those treated with fluconazole. Whether these data show that voriconazole is potentially more hepatotoxic than other azoles, as suggested by Potoski and Brown [1], remains to be established. Patients with invasive aspergillosis or febrile neutropenia demonstrated higher frequencies of abnormal LFT results overall than did patients with esophageal candidiasis. However, no differences were found in the frequency of such abnormalities between patients treated with voriconazole and those receiving amphotericin B formulations. This observation occurred even though the duration of exposure to voriconazole was greater than the duration of exposure to amphotericin B. These data suggest that, among patients who have multiple comorbidities and who receive multiple medications, the frequency of abnormal LFT results may depend on factors other than the type of antifungal agent. Association between clinically significant abnormal liver function test (LFT) results, irrespective of whether baseline values were normal or abnormal. To identify any possible relationship between plasma concentrations of voriconazole and increases in LFT values a longitudinal logistic regression analysis was conducted [5, 6]. With use of 2925 random plasma samples obtained from 1053 patients during 10 clinical trials, 7-day mean concentrations of voriconazole were calculated. The occurrence of abnormal LFT results and associated ORs are shown in table 2. The maximum rates of abnormal LFT results were observed at the highest plasma concentrations. This analysis showed that abnormal levels of aspartate aminotransferase (AST), alkaline phosphatase (ALP), and bilirubin, but not alanine aminotransferase (ALT), were significantly, but weakly, associated with plasma concentrations of voriconazole. The model predicted a 7%–17% increase in the odds of an abnormal ALT, AST, ALP, or bilirubin level for every 1 μg/mL increase in the plasma concentration of voriconazole. However, these investigations did not identify a threshold plasma concentration above which the risk of abnormal LFT results was increased when compared with lower plasma concentrations. For example, the risk of developing of an abnormal ALP level with a plasma concentration of voriconazole of 7 μg/mL is estimated to be ∼8%. Pototski and Brown [1] speculated that there might be an association between a plasma concentration of voriconazole of >6 μg/mL and abnormal liver function. If a decision to modify therapy to avoid toxicity were made solely on the basis of voriconazole concentrations, then, in this example, 11 of 12 patients would have their treatment discontinued or reduced unnecessarily. Maximum rate and OR for occurrence of abnormal liver function test (LFT) values per plasma voriconazole concentration increase of 1 μg/mL. Furthermore, a subset of 1134 samples obtained from 208 patients with aspergillosis and 762 samples obtained from 268 patients with candidiasis was used for analysis of the relationship between the plasma concentration of voriconazole and the drug's efficacy (data on file at Pfizer). No correlation between plasma concentration and efficacy was found. As for abnormal LFT results, the model also failed to identify upper and lower threshold plasma concentrations that are predictive of therapeutic failure. A validated assay to determine plasma concentrations of voriconazole has been developed and is available for research purposes. However, we disagree that voriconazole therapy should be guided by the individual plasma level measurements, because plasma concentrations of voriconazole are not predictive of abnormal LFT findings or treatment outcomes. It is unlikely that therapeutic drug monitoring will add any value over clinical judgement and diligent monitoring of LFT findings. At this time, no definite guidance can be provided for the interpretation of plasma concentrations of voriconazole.