|Home | About | Journals | Submit | Contact Us | Français|
Invasive fungal infections (IFI) are a major cause of morbidity and mortality among both solid organ transplant (SOT) and hematopoietic stem cell transplant (HSCT) recipients. Candida is the most common cause of IFI in SOT recipients and the second most common cause of IFI in HSCT recipients. We determined susceptibilities to fluconazole, voriconazole, itraconazole, posaconazole, amphotericin B, and caspofungin for 383 invasive Candida sp. isolates from SOT and HSCT recipients enrolled in the Transplant-Associated Infection Surveillance Network and correlated these results to clinical data. Fluconazole resistance in C. albicans, C. tropicalis, and C. parapsilosis isolates was low (1%), but the high percentage of C. glabrata and C. krusei isolates within this group of patients increased the overall percentage of fluconazole resistance to 16%. Voriconazole resistance was 3% overall but was 8% among C. glabrata isolates. On multivariable analysis, among HSCT recipients fluconazole nonsusceptibility was independently associated with C. glabrata, non-Hodgkin's lymphoma, cytomegalovirus (CMV) antigenemia, diabetes active at the time of the IFI, and any prior amphotericin B use; among SOT recipients, fluconazole nonsusceptibility was independently associated with any fluconazole use in the 3 months prior to the IFI, C. glabrata, ganciclovir use in the 3 months prior to the IFI, diabetes acquired since the transplant, and gender.
Both solid organ transplant (SOT) and hematopoietic stem cell transplant (HSCT) recipients are at high risk for invasive fungal infections (IFI) due to risk factors including long-term central venous catheter use, immunosuppression, and cytopenia, as well as therapy given to prevent graft-versus-host disease or rejection (5, 28). Candida species are the leading cause of invasive fungal infections in SOT patients, causing 53% of fungal infections, and the second leading cause of invasive fungal infections in HSCT patients, causing 28% of infections (13, 20).
In the United States, Candida albicans has been the most common cause of invasive candidiasis, and most isolates of C. albicans have been susceptible to azole antifungal drugs (21). Over the last 2 decades there has been a steady increase in the proportion of non-C. albicans Candida infections and a concomitant decrease in the proportion of infections caused by C. albicans (9, 12, 22, 30). This is a matter of concern because Candida glabrata and Candida parapsilosis, species with decreased in vitro susceptibilities to azole and echinocandin antifungal drugs, respectively, are increasing (22, 23, 25). Although the incidence rate of Candida krusei does not seem to be increasing, this fluconazole-resistant species seems to have been disproportionately represented among bone marrow transplant patients in several studies (1, 17, 24, 34).
The Transplant-Associated Infections Surveillance Network (TRANSNET), a consortium of 23 U.S. academic transplant centers, was established to conduct prospective surveillance to determine the burden of invasive fungal infections among transplant recipients (13, 20). It is important to monitor for the emergence of antifungal-resistant isolates and increases in resistant species within the vulnerable transplant population. In this study, we examined the antifungal susceptibility profiles of 383 incident invasive isolates of Candida from SOT and HSCT patients in the TRANSNET cohort and correlated clinical variables with isolate nonsusceptibility.
TRANSNET, a network of 23 U.S. transplant centers, conducted prospective surveillance for invasive fungal infections and has been described in detail elsewhere (13, 20). Briefly, participating centers enrolled and performed follow-up on 16,808 SOT and 15,820 HSCT recipients between March 2001 and March 2006. In all proven or probable cases of invasive fungal infection, as defined by modified European Organization for Research and Treatment of Cancer/Mycoses Study Group (EORTC/MSG) criteria (4), clinical data were collected by trained personnel at each site. All proven or probable cases of Candida infection with available isolates were selected for this analysis.
Available isolates were sent to the Fungus Reference Unit at the Centers for Disease Control and Prevention (CDC), Atlanta, GA, for confirmation of species and antifungal susceptibility testing. Isolate purity was confirmed using CHROMagar Candida (BBL, Franklin Lakes, NJ). Species identification was confirmed using API 20C AUX (bioMérieux, Durham, NC), and when API did not give excellent or very good results, DNA sequencing of the D1/D2 region of the rDNA (15) was used. Only the first isolate from each patient was used for analysis.
Prior to antifungal susceptibility testing, each isolate was subcultured twice on Sabouraud dextrose agar plates to ensure purity and optimal growth. One isolate per patient was tested. Broth microdilution MIC testing was performed as outlined in Clinical and Laboratory Standards Institute (CLSI) document M27-A2 (18) and as previously described (9) for fluconazole, itraconazole, and voriconazole. MICs of amphotericin B, caspofungin, and posaconazole were determined by Etest (AB Biodisk, Solna, Sweden) according to the manufacturer's instructions. Plates were incubated at 35°C for 24 h for caspofungin or 48 h for amphotericin B and posaconazole. To confirm the Etest results for posaconazole, broth microdilution was repeated for any isolate with an initial posaconazole Etest MIC of ≥1 μg/ml and for all C. glabrata isolates. All final posaconazole MICs of ≥1 μg/ml were the values derived from broth microdilution. Quality control was performed on each day of testing using C. krusei strain ATCC 6258 and C. parapsilosis strain ATCC 22019.
Isolates were categorized as susceptible, susceptible-dose dependent, or resistant based on established MIC cutoff values for fluconazole, itraconazole, and voriconazole and as susceptible or nonsusceptible for caspofungin (6). Fluconazole nonsusceptibility in the bivariate and multivariate analyses was defined as an MIC of >8 μg/ml.
For each transplant type, clinical information for case patients with susceptible isolates was compared to information for nonsusceptible isolates in order to identify factors associated with nonsusceptibility. Nine isolates were excluded from the modeling because no clinical information was available. Bivariate logistic regression was used to identify potential variables to evaluate for inclusion in a multivariate model for fluconazole nonsusceptibility.
For the multivariate model, variables with a P value of ≤0.2 were considered in addition to basic demographic variables. The 0.2 cutoff value was chosen to limit the number of potential variables for the multivariate model without being too restrictive in our selection criteria. Selected immunosuppressive medications, regardless of P value, were also considered for the model due to increased risk of infection with use. Antifungal medications, regardless of P value, were also considered for the model. Separate models were built for HSCT and SOT to reduce potential confounding due to transplant type. Logistic regression was used to build the multivariate models for fluconazole nonsusceptibility. The final multivariate model was chosen using the score selection method after evaluating for interaction and colinearity.
Differences between geometric means were calculated using Student's two-sample t test assuming unequal variances. Logistic regression for bivariate and multivariable analyses was conducted using SAS 9.2 (SAS Institute, Cary, NC).
A total of 915 proven and probable cases of invasive candidiasis were identified in TRANSNET (13, 20). Candida albicans was the most frequent species found (n = 350, 38% of cases), followed by C. glabrata (n = 250, 27%), C. parapsilosis (n = 94, 10%), Candida tropicalis (n = 51, 6%), and C. krusei (n = 30, 3%). Among HSCT patients, the most frequent species was C. glabrata (n = 92, 33% of cases), followed by C. albicans (n = 55, 20%), C. parapsilosis (n = 39, 14%), C. tropicalis (n = 23, 8%), and C. krusei (n = 17, 6%). In SOT recipients, C. albicans was the most frequent species (n = 295, 46%), followed by C. glabrata (n = 158, 25%), C. parapsilosis (n = 55, 9%), C. tropicalis (n = 28, 4%), and C. krusei (n = 13, 2%).
There were notable differences in the distribution of species between the SOT and HSCT populations. While 46% of isolates from SOT recipients were C. albicans, only 20% of isolates from HSCT recipients were C. albicans. In the HSCT population, 39% of isolates were C. glabrata and C. krusei, while only 27% of isolates from the SOT population were these species.
A total of 383 isolates were available from the 915 cases for susceptibility testing and species confirmation at CDC. Isolates were received from 20 of the 23 transplant centers. Four of the centers contributed a combined total of 58% of the isolates, with 17%, 17%, 13%, and 11% coming from those centers individually. The remaining isolates were spread somewhat evenly among the remaining 16 centers. The distribution of species among isolates received at CDC was not significantly different from the overall TRANSNET species distribution, with the exception that CDC received slightly more cases of C. krusei, with most coming from isolates which had no previous species identification performed (Table 1). Isolates were from blood (n = 238), peritoneal fluid (n = 41), abdominal fluid (n = 31), pleural fluid (n = 9), abscess (n = 6), liver (n = 6), ascites fluid (n = 6), lung (n = 5), bile (n = 5), tissue (n = 5), pancreas (n = 4), kidney (n = 3), bone (n = 3), mediastinal fluid (n = 2), wound fluid (n = 2), and other (n = 17).
The antifungal susceptibility testing results for all isolates are summarized in Table 2. All isolates were susceptible to caspofungin. Excluding isolates of C. krusei, 10% of isolates were resistant to fluconazole, including 23% of C. glabrata isolates (Table 2). When C. krusei isolates were added, the overall resistance was 16%. Among the C. albicans, C. parapsilosis, and C. tropicalis isolates (58% of the total isolates tested), resistance to fluconazole was only 1%. There was no association between any center and the proportion of fluconazole-resistant isolates received.
The overall resistance to itraconazole was 17%, again due primarily to the prevalence of C. glabrata isolates, and that among C. albicans, C. parapsilosis, and C. tropicalis was again around 1%. Voriconazole resistance was observed in 3% of isolates overall: two isolates of C. albicans, one isolate of C. tropicalis, and nine C. glabrata isolates. Voriconazole resistance was lower than fluconazole resistance among C. glabrata isolates, with 8% resistant and 76% fully susceptible to voriconazole. All of the C. glabrata isolates that were resistant to voriconazole were also resistant to itraconazole and to fluconazole. They also had posaconazole MIC values at or above the average achievable serum levels at 400 mg twice a day (BID) (7). All of the C. krusei isolates had voriconazole MIC values of ≤1 μg/ml, which is interpreted as fully susceptible.
There are no established breakpoints for resistance to amphotericin B. The overall MIC90 was 2.0 μg/ml. Four isolates of C. glabrata, 1 isolate of C. guilliermondii, and 13 isolates of C. krusei displayed MIC values of ≥2 μg/ml, a value at which treatment with amphotericin B would not be recommended.
The geometric mean MIC values for HSCT cases were significantly higher than those for SOT cases for fluconazole (4.099 versus 1.206 μg/ml, P < 0.001), voriconazole (0.132 versus 0.000 μg/ml, P = 0.002), itraconazole (0.218 versus 0.115 μg/ml, P = 0.007), and caspofungin (0.150 versus 0.094 μg/ml, P < 0.001).
We also found higher geometric mean fluconazole MIC values among patients who had previously received fluconazole prophylaxis (Table 3). This association was seen among all 383 isolates from all transplant types (7.207 versus 1.238 μg/ml, P < 0.0001), among all 119 isolates from HSCT recipients (9.190 versus 2.403 μg/ml, P = 0.0013), and among all 264 isolates from SOT recipients (4.876 versus 1.020 μg/ml, P = 0.0002). Among the 119 C. glabrata cases, higher fluconazole MIC values were associated with having received fluconazole prophylaxis (geometric means of 19.870 versus 10.556 μg/ml, P = 0.0133).
On bivariate analysis, fluconazole nonsusceptibility (MIC of >8 μg/ml) was significantly associated with having received an HSCT compared to an SOT (odds ratio [OR], 4.02; 95% confidence interval [CI], 2.42 to 6.66). This strong association with transplant type necessitated separate analyses for each transplant type.
Bivariate analyses for nonsusceptibility among HSCT and SOT recipients are shown in Table 4. Among the HSCT recipients, non-Hodgkin's lymphoma (OR, 2.52; 95% CI, 0.93 to 6.77), cytomegalovirus (CMV) coinfection (OR, 2.81; 95% CI, 1.05 to 7.48), diabetes active at the time of IFI (OR, 3.65; 95% CI, 1.45 to 9.19), and prophylactic antifungal use (OR, 3.49; 95% CI, 1.49 to 8.19) were associated with fluconazole nonsusceptibility. Among SOT recipients, female gender (OR, 3.14; 95% CI, 1.51 to 6.54), diabetes acquired since the transplant (OR, 2.74; 95% CI, 1.29 to 5.84), and prophylactic antifungal use (OR, 3.58; 95%, CI 1.63 to 7.90) were significantly associated with fluconazole nonsusceptibility.
On multivariate modeling for fluconazole nonsusceptibility among HSCT recipients, prior fluconazole use trended toward statistical significance (adjusted odds ratio [aOR], 2.66; 95% CI, 0.93 to 7.62) (Table 5). Having C. glabrata was significantly associated with fluconazole nonsusceptibility (aOR, 10.47; 95% CI, 3.58 to 30.62). Other significant variables included non-Hodgkin's lymphoma, CMV antigenemia, diabetes active at the time of the Candida infection, and any amphotericin B use prior to Candida infection. The model was adjusted for gender, race, and age.
On multivariate modeling for fluconazole nonsusceptibility among SOT recipients, fluconazole use in the 3 months prior to the infection was significantly associated with fluconazole nonsusceptibility (aOR, 2.65; 95% CI, 1.17 to 5.99). Candida glabrata was again strongly associated with fluconazole nonsusceptibility (aOR, 4.70; 95% CI, 2.08 to 10.58). Other significant variables included ganciclovir use in the 3 months prior to the IFI, diabetes acquired posttransplant, and gender. The model was adjusted for race and age.
In this analysis of the largest collection of invasive Candida isolates from transplant recipients to date, we demonstrate a high prevalence of fluconazole resistance (16%) when considering both resistant isolates and intrinsically resistant species.
The species distribution within TRANSNET reflects the changes in the epidemiology of candidiasis described over the past 2 decades, where the proportion of invasive C. albicans infections has declined consistent with an increase in the proportion of infections due to non-C. albicans species, especially C. glabrata (2, 9, 12, 22). In our study, the lower proportion of C. albicans was more evident among HSCT patients, with C. glabrata (34%) more common than C. albicans (19%). These data are consistent with PATH Alliance registry findings for U.S. HSCT patients (19), which showed that among the 20 invasive Candida isolates from HSCT patients in that study, C. glabrata was the most frequent (35%), followed by C. parapsilosis and C. tropicalis (30% each), C. albicans (25%), and C. krusei (10%).
We also found that the species distributions were markedly different for HSCT and SOT recipients. These differences in species distribution may be due to dissimilar antifungal use and prophylaxis practices, with a higher proportion of HSCT patients than of SOT patients in our study receiving antifungal prophylaxis (63% versus 15%). Multiple studies have suggested that the use of azoles can drive the species distribution within a population toward species with intrinsic resistance (1, 8, 10, 16, 26, 33, 34). Still, other factors may be contributing to this change in species distribution (27, 32), and a few studies have shown no increase in the incidence of azole-resistant Candida species following azole prophylaxis (14, 29). At least one report recorded an increase in C. krusei among immunocompromised patients who had not been treated with fluconazole (11).
Our data demonstrate statistically significantly higher MIC values for C. glabrata among patients who received fluconazole prophylaxis. Alexander and coworkers (3) found that five HSCT patients who had previous exposure to fluconazole developed infections with C. glabrata isolates with high MIC values to both fluconazole and voriconazole. Our results were similar. Of the 25 C. glabrata cases with MIC values of ≥2 μg/ml (susceptible-dose dependent to resistant) to voriconazole, 17 patients (68%) had received prior fluconazole prophylaxis. Of note is that any amphotericin B use prior to Candida infection was also associated with fluconazole nonsusceptibility in HSCT patients. This may be indicative of the underlying susceptibility of the patient to IFIs in general, and in an HSCT patient if the IFI was Candida, it was likely C. glabrata.
Our study provided further evidence that any prior fluconazole exposure was associated with fluconazole nonsusceptibility. In addition to higher geometric mean MICs among those with prior fluconazole exposure, prior fluconazole use in SOT patients was independently associated with fluconazole nonsusceptibility when analyzed by a multivariate model; this association was not as strong among HSCT patients. CMV antigenemia was associated with fluconazole nonsusceptibility among HSCT patients only, as has been noted previously (17). However, it is possible that some of these associations represent colinearity with other conditions that were not captured by our data collection form. For example, we found that diabetes was significantly associated with nonsusceptibility for both types of transplants. This relationship may reflect the higher risk of infection among transplant recipients receiving higher doses of corticosteroids (and therefore corticosteroid-induced diabetes). However, diabetes has not been frequently described as a risk factor for resistant Candida, and this deserves further study.
Other studies have demonstrated higher Candida MIC values among patients receiving fluconazole prophylaxis. A recent study examined data from patients who underwent allogeneic HSCT or who had acute myelogenous leukemia and received prophylactic posaconazole, itraconazole, or fluconazole (16). The number of patients with C. albicans colonization decreased with antifungal therapy in all arms of the study, while the proportion of patients with C. glabrata colonization increased in the posaconazole and itraconazole arms and the proportion of patients with C. krusei colonization increased in the fluconazole arm. Perhaps their most insidious finding was that in 40% of the C. glabrata-colonized patients, the MICs for the isolates increased more than 4-fold during the course of prophylactic therapy. Similarly, Trifilio and colleagues (31) reported that of the six breakthrough Candida infections in HSCT patients receiving voriconazole prophylaxis, five of the isolates were C. glabrata and one was C. krusei. Although susceptibility testing was not performed, the authors noted that all of the infections occurred among inpatients with voriconazole trough levels of <2 μg/ml.
This study had several limitations. While TRANSNET captured 10 to 20% of all transplants occurring in the United States during the surveillance period, it may not be representative of the entire U.S. transplant population. Additionally, not all of the participating centers sent isolates to CDC in a proportional manner. While the isolates studied at CDC did not largely differ from the overall cohort with respect to transplant type and antifungal use, the species identification was not reported in a substantial number of cases, so an exact correlation cannot be obtained and there is a possibility of selection bias. Another limitation is the risk of a spurious significant finding from the multivariate modeling due to the small sample size and a large number of variables considered in the model selection process. A number of nonsignificant variables from the bivariate analysis were included in the multivariate modeling to control for factors believed to influence antifungal susceptibility, increasing the number of variables considered during the model selection process. Finally, Etest is a nonreference method for susceptibility testing.
Fluconazole nonsusceptibility does seem to be prevalent in the overall transplant population. However, the problem relates not to individual isolates of multiple species demonstrating resistance but rather to the high frequency of C. glabrata and C. krusei. Based on the results of this study, susceptibility testing of C. glabrata isolates would be recommended prior to treatment with azoles. The rates of resistance to fluconazole and voriconazole for C. albicans and C. parapsilosis were exceedingly low (1%) despite widespread use of antifungal prophylaxis, and therefore routine antifungal susceptibility testing of these two species is not recommended. However, susceptibility testing should still be considered for breakthrough during prophylaxis or for other instances when clinical judgment would dictate usefulness, regardless of the species. Our results also suggest that susceptibility testing of isolates from patients responding to caspofungin therapy may not be warranted at this time.
We thank Mary Brandt for her extremely helpful critical reading of the manuscript and Kathleen Wannemuehler for help with the data collection.
TRANSNET was sponsored by Astellas, Pfizer, Merck, and Schering-Plough.
Potential conflicts of interest are as follows: P.G.P. receives research support from and is an ad hoc advisor for Merck, Pfizer, and Astellas; J.W.B. received research support from Pfizer and is a consultant/advisor for Merck and Pfizer; J.I.I. received honoraria for speaker activities from Astellas, Cubist, and Pfizer; D.R.A. has received grants from and is a consultant for Merck, Astellas, and Pfizer; and C.A.K. receives research support from Merck and chairs a data adjudication committee for Pfizer. All other authors have no conflicts.
The findings and conclusions of this article are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention.
Published ahead of print on 11 May 2011.