Should resistance to azole antifungals in vitro be interpreted as predicting clinical non-response?

Published data relating clinical treatment outcomes to susceptibility of Candida species in vitro for the triazole antifungal agents fluconazole and itraconazole show a clear association between rates of treatment failure and rising minimal inhibitory concentrations for the infecting fungal isolate. However, more than 50% of patients infected with an isolate 'resistant' to a triazole by NCCLS breakpoint criteria respond successfully to treatment with the triazole. Data for antibacterial agents similarly show that the association between resistance in vitro and treatment failure in vivo is far less than perfect. Susceptibility testing therefore falls into a category similar to that of weather forecasting. Despite good test standardization and sophisticated technology, the forecasts successfully predict trends, but cannot accurately foresee temperatures or levels of precipitation at a specific time in a specific location.