|Home | About | Journals | Submit | Contact Us | Français|
We prospectively determined the antifungal susceptibility of yeast isolates causing fungemia using the Etest on direct blood samples (195 prospectively collected and 133 laboratory prepared). We compared the Etest direct (24 h of incubation) with CLSI M27-A3 and the standard Etest methodologies for fluconazole, voriconazole, posaconazole, isavuconazole, caspofungin, and amphotericin B. Strains were classified as susceptible, resistant, or nonsusceptible using CLSI breakpoints (voriconazole breakpoints were used for posaconazole and isavuconazole). Categorical errors between Etest direct and CLSI M27-A3 for azoles were mostly minor. No errors were detected for caspofungin, and high percentages of major errors were detected for amphotericin B. For the azoles, false susceptibility (very major errors) was found in only two (0.6%) isolates (Candida tropicalis and C. glabrata). False resistance (major errors) was detected in 46 (14%) isolates for the three azoles (in 23 [7%] after excluding posaconazole). Etest direct of posaconazole yielded a higher number of major errors than the remaining azoles, especially for C. glabrata, Candida spp., and other yeasts. Excluding C. glabrata, Candida spp., and other yeasts, the remaining species did not yield major errors. Etest direct for fluconazole, voriconazole, isavuconazole, and caspofungin shows potential as an alternative to the CLSI M27-A3 procedure for performing rapid antifungal susceptibility tests on yeast isolates from patients with fungemia. Etest direct is a useful tool to screen for the presence of azole-resistant and caspofungin-nonsusceptible strains.
The incidence of fungemia continues to rise in many institutions throughout the world, and Candida is one of the leading pathogens isolated from blood (15, 28). Amphotericin B and fluconazole have been widely used for the treatment of fungemia. However, newly licensed antifungal agents (voriconazole, posaconazole, and the echinocandins) and other azoles currently under investigation (isavuconazole) have expanded the antifungal armamentarium.
The mortality rate of fungemia remains high (30%) and is clearly correlated with delayed initiation of effective antifungal therapy (11, 17). Antifungal therapy is considered inappropriate when it is omitted, when the agent administered has no antifungal activity against the infecting organism, or when its serum concentrations are subtherapeutic.
A growing proportion of Candida isolates obtained from blood samples have reduced antifungal susceptibility to fluconazole and other antifungal agents (14, 25). Patients with candidemia caused by Candida strains with high MICs for fluconazole or voriconazole and echinocandins can have a worse prognosis (22-24). Consequently, systematic use of empirical antifungal agents with broad-spectrum in vitro activity has led to considerable increases in the number of adverse events and in the cost of treatment (2).
The combination of an increasing number of antifungal-resistant isolates and the cost of the new antifungal agents makes antifungal susceptibility testing a necessity. The reference antifungal susceptibility testing method for yeasts is the Clinical and Laboratory Standards Institute (CLSI; formerly NCCLS) testing standard M27-A3. However, this method requires pure-culture isolates, and results of antifungal susceptibility testing are not available until 48 to 72 h after the isolation of fungi in blood.
The Etest performed directly on blood samples may expedite antifungal testing and provide results in 24 h. Our group has previously demonstrated that the Etest performed directly on samples from the lower respiratory tract is a rapid and accurate procedure for antimicrobial susceptibility testing of bacteria in patients with ventilator-associated pneumonia (4, 5). We compared the results of the Etest performed directly on positive blood cultures with yeasts grown in Bactec blood bottles with the results of CLSI M27-A3 in isolates from patients with fungemia and blood samples generated from previously characterized isolates.
(This study was partially presented at the 20th Conference of the European Congress of Clinical Microbiology and Infectious Diseases [ECCMID] in Vienna, Austria, 2010 [abstract no. P-838] [13a].)
The study was carried out prospectively from February 2007 to July 2009 in a 1,750-bed tertiary hospital (Hospital General Universitario Gregorio Marañón, Madrid, Spain) serving a population of 715,000 inhabitants. Blood samples from patients with fungemia were inoculated in Bactec blood culture bottles (Becton Dickinson Microbiology Systems, Cockeysville, MD) and incubated in the automated Bactec NR9240 instrument for no fewer than 5 days.
An episode of fungemia was defined as the isolation of yeasts in Bactec blood culture bottles. In patients with multiple blood culture sets containing yeast isolates, episodes were defined as follows: (i) a different yeast species isolated in subsequent blood culture sets or (ii) the same species isolated 1 month after the last blood sample was drawn. According to this definition, a total of 200 episodes of fungemia occurred in 194 patients.
Bactec bottles with a Gram stain revealing elements compatible with yeasts were selected. Only one bottle from the first blood culture set of each episode (n = 200) was selected. The distribution of microorganisms isolated was as follows: Candida albicans (n = 82), Candida parapsilosis (n = 64), Candida glabrata (n = 20), Candida tropicalis (n = 15), Pichia guilliermondii (formerly Candida guilliermondii) (n = 3), C. albicans plus C. parapsilosis (n = 3), C. albicans plus C. glabrata (n = 2), Rhodotorula mucilaginosa (n = 2), Candida dubliniensis (n = 2), Candida krusei (n = 1), Arxula adeninivorans (n = 1), Cryptococcus neoformans var. neoformans (n = 1), Kluyveromyces marxianus (formerly Candida kefyr) (n = 1), Saccharomyces cerevisiae (n = 1), Trichosporon inkin (n = 1), and Trichosporon mucoides (n = 1). All isolates were subcultured and stored at −70°C for antifungal susceptibility testing (CLSI and Etest standard procedures).
Due to the low number of episodes caused by species of Candida with reduced susceptibility to azoles or by non-Candida yeasts, we inoculated Bactec bottles with blood from nonfungemic/bacteremic patients with the following strains: C. glabrata (n = 47), C. krusei (n = 30), C. neoformans var. grubii (n = 15), C. neoformans var. neoformans (n = 13), S. cerevisiae (n = 14), Dipodascus capitatus (formerly Blastoschizomyces capitatus) (n = 9), T. mucoides (n = 2), C. tropicalis (n = 2) resistant to multiple azoles (fluconazole, voriconazole, posaconazole, and isavuconazole), and Cryptococcus gattii (n = 1).
In each case, a 0.5-ml volume of a suspension (0.5 McFarland standard) of each strain was inoculated into the bottles, which were reincubated until growth of yeast was detected. To estimate the number of yeast cells per ml of the 0.5 McFarland suspensions, we chose 10 strains of different species and counted the number of cells using a Neubauer chamber (Table (Table1).1). If the Bactec bottle gave a positive result, we also counted the number of yeast cells using microscopy. A previous study showed that variations in the yeast inoculum in the Bactec bottle had no impact on the MIC (6).
All isolates were identified using culture characteristics and biochemical reactions by means of CHROMAgar (Tec-Laim, Madrid, Spain) and ID 32C (bioMérieux, Marcy-l'Etoile, France). Strains of S. cerevisiae, D. capitatus, Trichosporon spp., P. guilliermondii, A. adeninivorans, and K. marxianus were identified by sequencing the ITS1-5.8S-ITS2 region, as described by White et al. (27). C. neoformans and C. gattii strains were identified by means of amplified fragment length polymorphism (AFLP) fingerprint analysis.
ETdir was prepared by pouring 10 to 20 drops from Bactec bottles onto 150-mm RPMI 1640 agar plates supplemented with 2% glucose (Tec-Laim, Madrid, Spain). The sample was further streaked across the surface of the agar plates, which were allowed to dry for 15 min. Etest strips (fluconazole, voriconazole, posaconazole, isavuconazole, amphotericin B, and caspofungin) were placed onto the surface and incubated at 35°C for 24 h. ETdir MICs could be interpreted in 85% of isolates after 24 h of incubation. All Cryptococcus isolates, one C. albicans isolate, four C. glabrata isolates, one C. parapsilosis isolate, two R. mucilaginosa isolates, and four S. cerevisiae isolates required 48-h incubation due to poor growth. A further eight S. cerevisiae isolates had to be incubated for 6 days.
Fifteen (7.5%) of the 200 Bactec blood culture bottles contained the following bacterial copathogens: Enterococcus faecalis (n = 1), Klebsiella pneumoniae (n = 1), Lactobacillus acidophilus (n = 1), Staphylococcus aureus (n = 1), and coagulase-negative Staphylococcus spp. (n = 11). In four of these bottles (2%), bacterial growth (coagulase-negative Staphylococcus spp. [n = 3] and E. faecalis [n = 1]) interfered with interpretation of the ETdir MICs, and they were excluded from the analysis. A strain from another blood culture was not available and was also excluded. A total of 328 bottles (195 prospectively collected and 133 laboratory prepared) were further analyzed.
The antifungal drugs obtained as reagent-grade powders were isavuconazole (Basilea Pharmaceutica International Ltd., Basel, Switzerland), voriconazole and fluconazole (Pfizer Pharmaceutical Group, New York, NY), posaconazole (Schering-Plough, Kenilworth, NJ), amphotericin B (Sigma, Madrid, Spain), and caspofungin (Merck Research Laboratories, Rahway, NJ). Stock solutions of caspofungin and fluconazole were prepared in sterile distilled water; the remaining agents were prepared in dimethyl sulfoxide (Sigma, Madrid, Spain). The final concentrations of drugs in the trays were 0.015 to 16 μg/ml (posaconazole, isavuconazole, voriconazole, and amphotericin B), 0.125 to 128 μg/ml (fluconazole), and 0.008 to 8 μg/ml (caspofungin). The inoculated trays were incubated at 35°C and read at 24 h (caspofungin and fluconazole) and 48 h (azoles and amphotericin B). Incubation was prolonged for Cryptococcus spp. and R. mucilaginosa (72 h) or S. cerevisiae (3 to 6 days) until fungal growth made it possible to interpret the MIC endpoint, defined as the lowest concentration at which growth is inhibited by 50% (7). Quality control was ensured by testing C. krusei ATCC 6258 and C. parapsilosis ATCC 22019 isolates. All results were within the recommended CLSI limits.
The ETsd was performed using yeast suspensions adjusted to a 0.5 McFarland standard. A swab was dipped into the suspension and streaked across the surface of the agar plates, which were allowed to dry for 15 min. Etest strips of the six antifungal agents were placed onto the surface and further incubated at 35°C for 48 h. The MICs were read according to the manufacturer's instructions.
As the Etest strips contain a continuous gradient of antifungal drug, the MICs obtained by the ETdir and ETsd were increased to the concentration of the next 2-fold dilution matching the drug dilution scale used for the CLSI procedure. All the MICs obtained were further converted to log2 MICs.
The mean MIC of each antifungal (obtained by the three different methods) was calculated. We used the t test to compare the antifungal activity determined by each method. The alpha value was set at 0.05, and all P values were two-tailed.
ETdir was compared with CLSI M27-A3 and ETsd procedures, which are considered the gold standards. ETsd was also compared with CLSI M27-A3. Agreement between the methods was considered to be essential when the log2 MIC measured by each method was within ±2 or fewer than 2-fold dilutions of the next one (8, 13, 21).
In addition, strains and antifungal agents were compared to calculate categorical agreement using the CLSI M27-A3 breakpoints, as follows: voriconazole (≤1 μg/ml, susceptible; 2 μg/ml, susceptible-dose dependent; ≥4 μg/ml, resistant); fluconazole (≤8 μg/ml, susceptible; 16 to 32 μg/ml, susceptible-dose dependent; ≥64 μg/ml, resistant); caspofungin (≤2 μg/ml, susceptible; ≥4 μg/ml, nonsusceptible). In the absence of breakpoints for posaconazole and isavuconazole, the breakpoints for voriconazole were used. Strains with MICs of >1 μg/ml for amphotericin B were considered resistant (7). The procedures were considered to be in categorical agreement when they resulted in the same susceptibility category (e.g., susceptible or resistant). Errors were categorized as very major (ETdir indicated susceptible and gold standard indicated resistant), major (ETdir indicated resistant and gold standard indicated susceptible), or minor (there was a single categorical shift between two results, e.g., susceptible to susceptible-dose dependent) (8).
The overall antifungal activity (MIC90) of each antifungal agent determined by the ETdir, ETsd, and CLSI procedures was as follows: posaconazole (ETdir/ETsd/CLSI), >32/8/0.5; isavuconazole, 2/1/0.25; fluconazole, >256/>256/32; voriconazole, 1/0.5/0.5; amphotericin B, 4/1/4; and caspofungin, >32/>32/16. For posaconazole, isavuconazole, and fluconazole (trays containing this drug were interpreted after 24 h and 48 h of incubation), the mean MICs obtained by ETdir or ETsd were significantly higher than those obtained by CLSI (P < 0.05). For voriconazole, the mean MIC obtained by ETdir was significantly higher than that obtained by CLSI (P < 0.05). For amphotericin B, the mean MIC obtained by CLSI was higher than that obtained by ETsd and ETdir and the ETdir mean MIC was higher than that obtained by ETsd. For the remaining comparisons, including caspofungin, the differences did not reach statistical significance.
Table Table22 summarizes the essential agreement found between the three procedures. In general, the highest agreement was found between ETdir and ETsd. Table Table33 shows the percentage of strains included in each susceptibility category (susceptible, susceptible-dose dependent, resistant, or nonsusceptible) for each antifungal agent. Table Table44 summarizes the categorical agreement found between the three procedures compared.
For fluconazole, with few exceptions, essential agreement between the ETdir and CLSI was higher when MICs were determined after 48 h of incubation. When C. krusei was excluded from the analysis, the overall essential agreement found was as follows: ETdir versus CLSI, 66.3% and 72.6% (for MICs obtained after 24 h and 48 h of incubation, respectively); ETdir versus ETsd, 86.5%; and ETsd versus CLSI, 69.1% and 79.3% (for MICs obtained after 24 h and 48 h of incubation, respectively). Agreement was near 100% for yeast species without fluconazole breakpoints proposed (e.g., Cryptococcus). Overall, CLSI tended to classify a higher number of isolates as susceptible. However, the percentages observed with the three methods were comparable for species showing low azole resistance (C. albicans, C. parapsilosis, and C. tropicalis). For fluconazole, comparisons of ETdir/ETsd with CLSI yielded a lower number of errors when MICs were read at 48 h. An 83.5% categorical agreement was found between ETdir and CLSI, and most errors were minor, especially for C. glabrata. We found only a minor error between ETdir and CLSI in one isolate of C. albicans (ETdir indicated susceptible and CLSI indicated susceptible-dose dependent). Very major errors for fluconazole, voriconazole, and posaconazole were detected in one isolate of C. tropicalis. Excluding a very major error found in C. krusei (ETdir MIC = 16 μg/ml; CLSI MIC = 64 μg/ml), the remaining major errors occurred only in C. glabrata or other yeasts. Taking into account the intrinsic fluconazole resistance of C. krusei, ETdir and ETsd detected fluconazole resistance in 90.3% of the isolates.
Among the azoles, ETdir and CLSI showed the highest essential agreement for voriconazole. This was particularly high for species showing reduced or intrinsic fluconazole resistance (C. glabrata and C. krusei). A categorical agreement of 94.3% was found between ETdir and CLSI. Errors for C. glabrata were mostly minor. Only two very major errors were detected in a fluconazole-resistant C. glabrata isolate (correctly classified by ETdir as resistant) and in another isolate of C. tropicalis. For C. parapsilosis, errors were minor in only one isolate (ETdir indicated susceptible-dose dependent and CLSI indicated susceptible).
Essential agreement between ETdir and CLSI was especially poor for posaconazole, with C. glabrata showing the lowest values. The essential agreement between ETdir and CLSI for isavuconazole was also low for C. glabrata. For this species, essential agreement for posaconazole and isavuconazole was moderately improved when the ETsd was chosen as the gold standard. The CLSI M27-A3 document has not proposed breakpoints for posaconazole or isavuconazole. Using the same breakpoints as for voriconazole, a high proportion of C. glabrata strains crept from the category of susceptible (CLSI) to resistant (ETdir or ETsd) for posaconazole. Errors for posaconazole and isavuconazole were found mostly for C. glabrata and several other yeasts.
The lowest agreement between ETdir and CLSI was found for amphotericin B, even when the two Etest procedures were compared. The overall essential agreement was 55.5%, and this rose above 80% for only three species. The CLSI M27-A3 document has not proposed breakpoints for amphotericin B. When susceptibility was defined as an MIC of ≤1 μg/ml (7), both Etest procedures tended to classify a higher number of strains as susceptible than the CLSI. Amphotericin B was the drug for which the ETdir showed the worst categorical agreement with the reference procedures (Etest standard or CLSI). The overall categorical agreement was improved when the ETsd was chosen as the gold standard (69.5% versus 55.5%), with the exception of C. glabrata (39.3%) and other yeasts (58.1%). Agreement was especially low for Cryptococcus spp.
Caspofungin was the drug for which ETdir and CLSI showed the highest values of essential and categorical agreement. Both methods presented high degrees of essential agreement (>90%), especially for C. glabrata and C. krusei. The three procedures were equivalent in terms of classifying isolates as susceptible. No errors were detected between ETdir and CLSI.
We used the ETdir on blood samples to determine the antifungal susceptibility of yeasts causing fungemia, and we compared this procedure with the reference microdilution procedure CLSI M27-A3 and the Etest standard. Although the ETdir overestimated azole resistance, it proved to be a rapid, easy, and reliable procedure for the preliminary determination of susceptibility to azoles and caspofungin of yeasts causing fungemia. ETdir is comparable to the CLSI M27-A3 procedure (7) and results can be obtained in less than 24 h after visualization of the yeast in the Gram stain.
Antifungal susceptibility testing of yeast isolates from blood is a standard of care in many tertiary hospitals. Clinicians can use this information to optimize the treatment of patients with fungemia (22, 23). The CLSI M27-A3 procedure is the reference method for antifungal susceptibility testing of yeasts, although it has some limitations, since it is time-consuming and labor-intensive and since results are not available until 24 to 48 h after pure-culture isolation of the fungal pathogen (7, 18). The Etest, an easy-to-perform procedure, can partially compensate for these limitations. The ETsd has proven to have a good correlation with the CLSI procedure (9, 16, 26). However, the MIC is read after 48 h of incubation and has been standardized using pure-culture isolates. ETdir on blood samples can provide results in 24 to 48 h. In 2001, Chang et al. compared the ETdir with the NCCLS macrodilution procedure in positive blood cultures for five antifungal agents. However, the recent incorporation of the echinocandins and the new triazoles, together with the development of new breakpoints, means that the procedure should be reevaluated (6).
We studied the ETdir results for posaconazole, isavuconazole, fluconazole, voriconazole, amphotericin B, and caspofungin. The highest essential agreement was found for caspofungin and the lowest for posaconazole and amphotericin B. In the categorical analysis, the highest correlation was also found for caspofungin and the lowest for amphotericin B. Errors between ETdir and CLSI for azoles were mostly minor.
False susceptibility (very major errors) was detected for azoles in only two (0.6%) isolates. One C. tropicalis isolate showed very major errors for fluconazole, posaconazole, and voriconazole but not for isavuconazole. C. tropicalis frequently exhibits azole trailing, and this may interfere with the interpretation of MICs (3). In this strain, the 24-hour CLSI MICs of the other azoles led to the same susceptibility category. The other very major error occurred in a C. glabrata isolate for which ETdir indicated susceptibility to voriconazole but resistance to fluconazole (CLSI indicated resistance for both agents), thus revealing the potential of the ETdir to detect fluconazole-resistant isolates of C. glabrata. Isavuconazole, a new triazole under phase III evaluations for the treatment of candidemia, has a favorable pharmacokinetic profile and an MIC distribution comparable to that of voriconazole (12, 13). When the voriconazole breakpoints were used, no very major errors were found between the results of ETdir and CLSI for isavuconazole. Both essential and categorical agreements between ETdir and CLSI for caspofungin were excellent, indicating the ability of the ETdir to correctly detect isolates that are susceptible or nonsusceptible to caspofungin, which is widely used for the treatment of candidemia. The MICs obtained by the ETdir for our strains were within the limit of very major errors (≤1.5%) proposed by the FDA (10), indicating the high potential of this procedure to screen for the presence of azole-resistant or caspofungin-nonsusceptible strains.
False resistance (major errors) was detected in 46 (14%) isolates for the three azoles (in 23 [7%] after posaconazole was excluded) and in 75 (22.9%) isolates for amphotericin B. The ETdir of posaconazole yielded a higher number of major errors than those of the remaining azoles, especially in the case of C. glabrata, Candida spp., and other yeasts. The results obtained using ETdir with posaconazole must be interpreted with caution, as there are no established breakpoints for this agent and this procedure overestimated antifungal resistance, especially in C. glabrata.
Excluding C. glabrata, Candida spp., and other yeasts, the remaining species did not yield major errors for azoles. However, the FDA proposed a limit of ≤3% for major errors. It must be emphasized that the ETdir overestimated resistance to azoles and that this may lead to the use of broad-spectrum antifungal agents for empirical treatment (10). However, it is important to note that the ETdir overestimated the azole resistance for C. glabrata, although it is still unknown whether azoles should be used for the treatment of episodes caused by this species. Few errors were found for highly prevalent species (i.e., C. albicans and C. parapsilosis); ETdir performed particularly well for caspofungin, which is widely used for the treatment of candidemia. Preliminary ETdir results should be confirmed using a standard procedure, such as CLSI M27-A3.
Antifungal susceptibility testing for amphotericin B is controversial. Microdilution procedures usually yield a narrower distribution of MICs than agar-based methods (19). The Etest has proven to be more sensitive and reliable than CLSI for detection of resistance to amphotericin B in Candida (20). We compared the ETdir of amphotericin B with the Etest standard procedure. Very major errors were detected in 9 (2.7%) isolates, and this percentage was above 1.5% for C. parapsilosis, C. krusei, C. tropicalis, and other yeasts. In the absence of consensus for an appropriate method for amphotericin B, these results must be interpreted with caution and the use of ETdir with amphotericin B cannot be recommended.
No major or very major errors were detected in C. albicans or C. parapsilosis, species that could account for 75% of cases of candidemia (1). However, a potential shortcoming of our study is the low number of azole-resistant or caspofungin-nonsusceptible strains of C. albicans, C. parapsilosis, and C. tropicalis.
We conclude that the ETdir for fluconazole, voriconazole, isavuconazole, and caspofungin has the potential to provide a rapid preliminary antifungal susceptibility result in yeast isolates from patients with fungemia in lieu of data obtained using the standard CLSI M27-A3 procedure. ETdir is a useful tool to screen for the presence of azoles and caspofungin-nonsusceptible strains. Future prospective evaluations of the use of the ETdir in the clinical setting are warranted to determine the role of this procedure in the optimization of the antifungal treatment of patients with fungemia.
We thank Thomas O'Boyle for editing and proofreading the article.
This study does not present any conflicts of interest for its authors.
This study was partially financed by grants from Basilea Pharmaceutica International Ltd., Basel, Switzerland. Etest strips of isavuconazole were kindly supplied by Basilea Pharmaceutica. Etest strips of amphotericin B, fluconazole, voriconazole, posaconazole, and caspofungin were kindly supplied by AB Biodisk (Solna, Sweden).
Jesús Guinea (CP09/00055) and Marta Torres-Narbona (CM08/00277) are contracted by the Fondo de Investigación Sanitaria (FIS).
Published ahead of print on 14 April 2010.