Breakthrough candidaemia in the era of broad-spectrum antifungal therapies.

We aimed to assess the characteristics, treatment, risk factors and outcome of patients with breakthrough candidaemia (BrC) in the era of broad-spectrum antifungal therapies. We carried out a multicentre study of hospitalized adults with candidaemia at six hospitals in three countries. BrC episodes were compared with the remaining episodes (non-BrC). Of 409 episodes of candidaemia, 37 (9%) were BrC. Among them, antifungal treatment was administered as prophylaxis in 26 severely immunosuppressed patients (70%) and as a fever-driven approach in 11 (30%). Candida albicans was significantly less common in patients with BrC (24% versus 46%, p 0.010) whereas Candida krusei was more frequent (16% versus 2.4%, p < 0.001). BrC was associated with infections caused by fluconazole non-susceptible isolates (50% versus 18%, p < 0.001). Candida albicans BrC was associated with previous fluconazole treatment whereas Candida parapsilosis candidaemia was mostly catheter-related and/or associated with previous echinocandin therapy. The empirical antifungal therapy was more often appropriate in the non-BrC group (57% versus 74%, p 0.055). No significant differences were found in outcomes (early and overall mortality: 11% versus 13% p 0.802 and 40% versus 40% p 0.954, respectively). Fluconazole non-susceptibility was independently associated with the risk of BrC (adjusted OR 5.57; 95% CI 1.45-21.37). In conclusion, BrC accounted for 9% of the episodes in our multicentre cohort. The Candida spp. isolated were different depending on the previous antifungal therapy: previous azole treatment was associated with fluconazole non-susceptible strains and previous echinocandin treatment was associated with BrC caused by C. parapsilosis. These results should be taken into account to improve the empirical treatment of BrC.