Patients with implanted medical devices have an increased risk of candidaemia (Douglas, 2003; Ramage et al., 2006; Imamura et al., 2008). Candida generates a biofilm structure protecting the yeast from host defence and antifungal agents (Kuhn & Ghannoum, 2004; Tumbarello et al., 2007; Al-Dhaheri & Douglas, 2008). The current clinical procedure is to remove the catheter (Douglas, 2003; Kuhn & Ghannoum, 2004; Percival et al., 2005; Seidler et al., 2006; Cateau et al., 2008). Coating, flushing or antimicrobial lock technique (ALT) with chemicals or drugs can reduce or eradicate catheter-associated biofilms of clinically relevant microorganisms (Percival et al., 2005; Cateau et al., 2008; Jacobson et al., 2008). High concentrations of liposomal amphotericin B (LAMB) using ALT virtually cleared the catheters in rabbits (Schinabeck et al., 2004). Venkatesh et al. (2009) reported additionally that N-acetylcystein, EDTA, ethanol and talactoferrin in combination with antibiotics used as catheter lock solutions have a synergistic effect on biofilms in catheters. The aim of this study was to test LAMB at minimal inhibitory concentration (MIC) in a Candida albicans biofilm catheter continuous steady flow model. Previous reports on antifungal treatment of biofilms in catheters are mostly restricted to blocking the catheter with high concentrations of antimicrobials. The novelty of this model is the continuous flow investigating catheters used in the clinic. With this model, we are able to somehow mimic the development of Candida biofilms in implanted Hickman catheters. In the literature, the reports on fungal biofilms and catheters are mostly limited to in vitro or in vivo infection models (Donlan, 2008).
Candida albicans ATCC #MYA-2876 blastospores were adjusted to 106 mL−1. First, antifungal susceptibility tests were performed following the guidelines of the Clinical Laboratory Standards Institute (2008) and the XTT test for Candida biofilms (Chandra et al., 2008). Hickman catheters were screwed onto a 50-mL syringe and placed into a perfusor. One catheter was installed in the flow (2 mL min−1) and pure media (RPMI+5% FCS) were passed for 48 h to obtain an uninfected and untreated control. Two catheters were infected by passing the yeast solution for 24 and 48 h to obtain infected, but untreated catheters with produced biofilm. The remaining nine catheters were infected and left to produce biofilm in the flow for 24 h and were treated afterwards by passing LAMB, fluconazole and caspofungin directly into the flow for another 24 h. All antifungal agents were used at a concentration of 0.5 μg mL−1. Three parts of the catheter were chosen randomly: near the tip, in the middle and near the end. These were compared with each other for a better reproducibility. The catheters were cut into discs and stained in a 1% safranin solution to observe the growth morphology. For scanning electron microscopy (SEM), the catheters were cut into pieces, bisected lengthwise, vaporized with graphite and placed onto the SEM adapter. For confocal scanning laser microscopy (CSLM) staining, the catheters were cut into discs and the polysaccharides were stained with concanavalin A-Alexafluor 488 conjugate (CAAF). All observations of the LAMB-, fluconazole- and caspofungin-treated catheters were compared with the control catheters visually: the images of the untreated catheters infected for 24 h were classified as intermediate (50%) and 48 h as mature biofilm (100%). The comparison with controls using percentages is not only performed for microscopy studies but also for metabolic dyes such as XTT. For XTT, the % reduction of metabolic activity is expressed in comparison with controls (Seneviratne et al., 2008). The CFUs of the catheter treated with LAMB were counted after a 24-h incubation, additionally.
All three antifungal drugs displayed good in vitro activity against planktonic C. albicans (0.25–0.5 μg mL−1). After 24-h biofilm, the MICs of each drug were increased by one dilution. After 48-h biofilm, caspofungin demonstrated the best activity, with 1 μg mL−1, followed by LAMB with 4 μg mL−1; fluconazole was not active, with MICs >16 μg mL−1. Figure 1 (left column) displays the morphology of C. albicans inside the catheters after safranin staining. Without the influence of antifungal agents, a wide dense and time-dependent growth of pseudohyphae was observed with an ensuing decrease in blastospore production. After 24-h biofilm formation and a 24-h treatment with LAMB, the growth of the hyphal network was reduced to 20% in comparison with the untreated control. Treating with fluconazole and caspofungin, the growth of the network was stopped and remained at the intermediate phase (50%). A morphologic change of the blastospores and the pseudohyphae was not observed. Figure 1 (middle column) displays the topographic surface of C. albicans biofilm (B) and the catheter (C) as an interface. The catheter surface without attached biofilm formation was clear with sparse attachment of blastospores and pseudohyphae. The surface of the infected, but untreated catheters showed 40% at 24 h, and 80% at 48 h, total covering. After 24-h biofilm formation and a 24-h treatment with LAMB, 20% of the surface was covered in biofilm. For caspofungin and fluconazole, the surface covering was 80% and therefore similar to the observations of the infected, but untreated catheter after 48 h. Despite LAMB, the surface of the biofilm was even and the fungus was completely packed in extracellular matrix (ECM). LAMB caused an uneven surface. Figure 1 (right column) shows the CSLM images with stained ECM for comparison of the thickness of the matrix. The uninfected and untreated catheter had a clear edge without produced matrix formations. The ECM of the infected, but untreated catheters had a thickness of 5–20 μm at 24 h and 10–150 μm at 48 h. After 24-h biofilm formation and a 24-h treatment with LAMB, the ECM was virtually cleared with 0 μm ECM. After 24-h biofilm formation and a 24-h treatment with fluconazole, the ECM thickness was comparable to the infected, but untreated catheter at 24 h with 10–25 μm; with caspofungin, the ECM thickness was comparable to the infected, but untreated catheter at 48 h with 10–130 μm (Fig. 1, right panel). The CFUs of the LAMB-treated catheter displayed a linear time-dependent reduction of viable Candida from an overgrown state at 0-h LAMB treatment over 394 CFUs at 2 h to 19 CFUs at 24 h, confirming the microscopic observations.
Lipid formulations of LAMB and the echinocandins exhibited activity against mature biofilms, while azoles failed to exert activity against mature Candida biofilms in vitro (Kuhn & Ghannoum, 2004). It is still controversial as to whether in vitro produced biofilms can reflect the in vivo situation. In vivo implanted catheters are rinsed, flushed or stay close to the systemic blood circulation. Candida can reach the mature biofilm state with a delay of 1 day in the liquid flow compared with in vitro tests (Chandra et al., 2008). In our study, liquid flow and delayed biofilm production seem not to affect the development of blastospore and pseudohyphae.
Schinabeck and colleagues infected surgically placed silicone catheters in New Zealand White rabbits with C. albicans, and used the ALT with 3 mg LAMB in 300 μL solution. SEM revealed abundant biofilm in the control and fluconazole groups, while the LAMB group was virtually cleared (Schinabeck et al., 2004). Mukherjee and colleagues determined the efficacy of amphotericin B lipid complex (ABLC) against C. albicans biofilms using a rabbit model of catheter-associated candidal biofilm. After lock treatment with 1.5 mg ABLC, the SEM analysis showed that while catheters retrieved from untreated control animals were overlaid with a thick biofilm, those treated with ABLC lock therapy showed only some debris with no fungal cells (Mukherjee et al., 2009). Nonetheless, our study, using a continuous flow with low concentrations for only 1 day, has a comparable effect to ALT blocking of the catheter with very high concentrations for 7 days. Cateau and colleagues reported recently that caspofungin (2 mg L−1) could represent a good candidate for the reduction of fungal biofilms associated with silicone medical devices, as part of an ALT. In this study, a fourfold lower concentration of all antifungal agents was used. This could explain why caspofungin was inefficient in the Cateaus report (Cateau et al., 2008). Medical device substrates such as denture strips, catheter disks or contact lenses have been investigated regarding their fungal metabolic activity by XTT or dry weight and their morphology using fluorescence microscopy, SEM and CSLM techniques (Chandra et al., 2008).
We were able to demonstrate that LAMB was effective at MICs against matured Candida biofilms in catheters by reducing the growth of blastospores, their morphology changes to pseudohyphae, the surface covering and the production of ECM. Whether developing Candida biofilms in implanted catheters can be eradicated by LAMB will be further investigated in a catheter rat model to address the clinically relevant question as to whether a surgically implanted and infected catheter should not be removed under all circumstances.