A single-centre 10-year experience with Candida bloodstream infections

1Department of Microbiology-Infectious Diseases, Hôpital Maisonneuve-Rosemont; 2Department of Microbiology and Immunology, Université de Montréal, Montreal; 3Department of Microbiology and Infectious Diseases, Université de Sherbrooke, Sherbrooke, Quebec Correspondence: Dr Michel Laverdiere, Department of Microbiology-Infectious Diseases, Hôpital Maisonneuve-Rosemont, 5415 boul de l’Assomption, Montreal, Quebec H1T 2M4. Telephone 514-252-3817, fax 514-252-3898, e-mail michel.laverdiere@umontreal.ca Received for publication May 12, 2008. Accepted October 1, 2008 Since the original observation by Krause et al (1) of the invasive potential of Candida species, a considerable number of studies have expanded our knowledge on Candida bloodstream infections (BSIs). In a large database of nosocomial BSIs in the United States, Candida species ranked fourth overall among etiological agents, and was the third most common cause of such infections in intensive care units (ICUs) (2). This reflects increasing numbers of immunocompromised patients, greater use of antibiotics, better diagnostic methods, frequent use of invasive procedures and improvements in life-support systems (3-8). Over the past 20 years, the epidemiology of candidemia has been considerably modified by the emergence of non-albicans Candida species, the occurrence of triazole-resistant species and the development of secondary resistance among previously sensitive isolates (4,5,9-11). Pfaller and Diekema (12) recently emphasized the importance of sustaining surveillance programs to OriginAl Article


MethODS
Hôpital Maisonneuve-Rosemont is a tertiary care academic hospital in Montreal, serving a population of approximately 500,000 inhabitants, with a large hematopoietic stem cell transplantation (HSCT) program since 1980, a kidney transplantation program, and all medical and surgical specialties.All candidemia episodes observed between August 1996 and July 2006, among patients 18 years of age or older, were retrieved from the microbiology laboratory records.Repetitive episodes, defined as the same species isolated from the same patient, were excluded.Different species isolates in the same patient were considered to be distinct episodes.Standardized forms were used to collect clinical data from hospital records.Patients' localization at the time of the diagnosis was categorized.Speciation and antifungal susceptibility testing of the recovered isolates were performed by means of germ tube, chlamydospore formation, Vitek YBC identification system (BioMérieux, Canada) and the Clinical and Laboratory Standards Institute (USA) broth microdilution susceptibility testing method (17).
Standard antifungal powders of amphotericin B, itraconazole (Sporanox, Janssen-Ortho Inc, Canada), ravuconazole (Bristol-Myers Squibb, Canada), fluconazole (Diflucan, Pfizer Canada), voriconazole (Vfend, Pfizer Canada Inc), posaconazole (Noxafil, Schering-Plough, USA), micafungin (Mycamine, Astellas Pharma Canada Inc), caspofungin (Cancidas, Merck Frosst Canada Ltd) and anidulafungin (Eraxis, Pfizer Canada Inc, Canada) were supplied from their respective manufacturers.Stock solutions were prepared according to manufacturers' specifications.Serial twofold dilutions were made in RPMI-1640 medium (Sigma Aldrich Canada) buffered to pH 7.0 with 0.165 M morpholinepropanesulfonic acid buffer (Sigma-Aldrich Canada) for all antibiotics.The final concentrations of the antifungal agents were 0.008 mg/L to 16 mg/L for amphotericin B, itraconazole, voriconazole, ravuconazole, posaconazole, micafungin, caspofungin and anidulafungin, and 0.25 mg/L to 256 mg/L for fluconazole.Drug-free and yeast-free controls were included.Trays were incubated in air at 35 o C, and the minimum inhibitory concentration (MIC) end points were visually read after 48 h of incubation.The MICs of amphotericin B were defined as the lowest concentration resulting in a complete inhibition of growth, while the MICs of all the other compounds were defined as the lowest concentration that resulted in a substantial reduction of growth (80% inhibition) compared with that of growth-control wells.Candida parapsilosis ATCC 22019 and Candida krusei ATCC 6258 were used for quality control.
Data were analysed with Stata 8.0 (StataCorp LP, USA).Medians were compared using the Kruskal-Wallis test.Proportions were compared using the c 2 test or Fisher's exact test when appropriate.Unconditional logistic regression was used for multivariate analysis.Models were built up sequentially, starting with the variable most strongly associated with the outcome and continuing until no other variable reached significance.When the final model was reached, each variable was dropped in turn to assess its effect.Different models were compared using the likelihood ratio test, keeping only variables significant at the P=0.05 level in the final model.

ReSuLtS
A total of 190 distinct episodes of candidemia in 185 patients were documented.Three patients were infected with two different Candida species on the same day, and two patients experienced two separate episodes of candidemia 10-day (C parapsilosis and Candida glabrata) and 11-day (C albicans and C krusei) intervals, respectively.The yearly number of episodes ranged from 11 to 26 (median of 18 episodes).The overall species distribution by study periods is described in Table 1.The proportion of C albicans over the 10-year observation period remained fairly stable (57% overall), and the higher proportion of C krusei candidemia episodes found between 2003 and 2006 were not statistically significant (13% compared with 8% between 1996 and 1999; OR 2.1; 95% CI 0.6 to 7.3).At least two different blood culture samples were drawn for 181 (95%) patients (median of four samples per patient; range two to 36 samples over one to 26 days).In 66 of 181 (36%) patients, only one sample was positive (transient candidemia).Overall, same episode blood cultures remained positive for two or more days in 82 (43%) episodes (median of five episodes; range two to 25 episodes).
Demographic characteristics, localization of patients and risk factors identified at the time of diagnosis are listed in Table 2.There were no significant differences between patients infected with C albicans strains and those infected with nonalbicans Candida species.Among HSCT recipients, 24% were infected with C albicans and 14% with non-albicans Candida species.The overall median age increased over the study period (48 years of age between 1996 and 1999, 56 years of age between 2000 and 2002, and 64 years of age between 2003 and 2006; P=0.02).Episodes were evenly distributed among patients in the hematology-oncology unit (including hematopoietic stem cells recipients), those in the ICU and those in the medical/ surgical wards, each contributing to approximately one-third of the episodes.However, the proportion of patients who were on the hematology-oncology unit at the time of their candidemia decreased from 41% between 1996 and 1999 to 19% between 2003 and 2006 (P=0.009).This trend was statistically significant for patients infected with C albicans (44% were on the Antifungal therapy was administered in 174 episodes (92%).Deoxycholate amphotericin B was used in 84 episodes (44%), a lipid-based amphotericin formulation in 26 (14%), fluconazole in 128 (67%), itraconazole in one (1%), caspofungin in 30 (16%) and anidulafungin in one (1%).Among the 16 patients who were not treated, nine died within two days before their blood culture positivity was known; in four others, the time to positivity of their blood cultures ranged from 20 h to 3.3 days; three patients had only one positive blood culture bottle (two with C albicans and one with Candida tropicalis), no known risk factor for Candida infection and survived.Death within 30  Comparative in vitro susceptibilities, along with the currently established Clinical and Laboratory Standards Institute (USA) interpretations criteria, are summarized in Table 3.All C krusei strains were considered resistant to fluconazole, irrespective of the MIC results.Among the other Candida species, nine (5%) isolates were resistant to fluconazole (MIC 64 mg/L or greater).Overall, the MICs required to inhibit the growth of 90% of organisms of fluconazole and voriconazole were lower than the breakpoint value for resistance.The MICs required to inhibit the growth of 90% of organisms of posaconazole and ravuconazole, for which breakpoints have not yet been established, were equally low at 1 mg/L.Itraconazole and fluconazole were less active against non-albicans Candida species taken together (n=81), than against C albicans species (only 26% of non-albicans Candida species isolates were susceptible to itraconazole [P<0.001] and 67% to fluconazole [P<0.001]).However, voriconazole activity against C albicans and non-albicans Candida species was identical (96%).For C albicans, fluconazole resistance remained fairly stable during the study period -6% between 1996 and 1999, 2% between 2000 and 2002, and 3% between 2003 and 2006.For other species, the fluconazole resistance rate was higher between 2003 and 2006 (36%) compared with 1996 and 1999 (24%) and 2000 and 2002 (21%).This difference was not statistically significant and is likely due to a higher number of C krusei isolated between 2003 and 2006.
Of the 26 (14%) fluconazole-resistant isolates (MIC 64 mg/L or greater), 10 (38%) were also resistant to itraconazole; cross-resistance or reduced susceptibilities (MIC 4 mg/L or greater) for the newer azoles (voriconazole, ravuconazole and posaconazole) were seen in only nine isolates (two C albicans, six C glabrata and one C tropicalis).There was no obvious epidemiological link among any of these nine isolates.All four fluconazole-resistant C albicans were isolated in hematology-oncology patients.
High MICs (2 mg/L or greater) for echinocandins were observed mainly with C parapsilosis.A tendency for a class effect was observed among these isolates with reduced susceptibility to  4. Age, hospital unit at the time of diagnosis and antifungal use within the past three months were significantly associated with fluconazole resistance.In multivariate analysis, being on the hematology-oncology unit at the time of diagnosis, compared with outpatients and patients on the medical or surgical wards (adjusted OR [AOR] 7.8; 95% CI 2.3 to 27.1; P=0.001) and having received fluconazole or itraconazole within the past three months (AOR 8.3; 95% CI 2.8 to 24.4; P<0.001) were significantly associated with resistance to fluconazole.Study period and duration of current hospital stay were not significantly associated with resistance to fluconazole.

DISCuSSIOn
The dominance of C albicans, which accounted for approximately 60% of our Candida BSI, is comparable with observations made in other Canadian centres (15,16,18,19).Although C glabrata was overall the second most frequent species (approximately 15%), variations were noted in the species ranking when analyzed by patients' localization.C parapsilosis was the second most common species in our outpatients' BSIs, reflecting the exogenous skin niche of this organism and its ability to form biofilms on indwelling intravascular catheters of patients managed at home (20)(21)(22).C krusei and C tropicalis ranked second (21%) and third (12%), respectively, in the hematology-oncology units patients, with C glabrata at a distant fourth place, with only 5% of the isolates.Each of these three species has been associated with hematological and solid tumours (23)(24)(25).Selective pressure from fluconazole prophylaxis has been incriminated in the emergence of C krusei and C glabrata among certain patient populations (11,23).In our institution, systematic prophylaxis has been used only in our HSCT recipients.Fluconazole has been our preferred prophylactic agent and was extensively used throughout the study period.The low prevalence of fluconazole-resistant C glabrata in our institution may explain why this species remained uncommon, although factors other than exposures to triazoles antifungals have been shown to select for these species (26).Approximately one-third of the episodes were noted in patients localized outside our ICU and hematology-oncology units at an increasing trend, which emphasizes the importance of maintaining a high level of clinical suspicion in the recognition of Candida BSIs in patient populations outside the traditional risk groups of compromised cancer and critical care patients.
During the latter years of our study, the frequency of candidemic patients with central venous catheters decreased, with a corresponding increase in the proportion of patients with a PICC line.This observation is likely attributable to a better accessibility in our centre to PICC line insertions.The insertion and nursing care conditions of these lines have been as rigorous as those applied to central venous lines.
Globally, the older triazoles (fluconazole and itraconazole) antifungals remain very active against the majority of our isolates.The echinocandins all had high in vitro activity against the various isolates, except for C parapsilosis.A class-effect tendency was observed with all three echinocandins, showing reduced activity against C parapsilosis, compared with their activity against the other Candida species.The clinical significance of this reduced in vitro activity appears negligible, with poor correlation between the MICs of echinocandin and treatment outcome, as reported by Kartsonis et al (27).However, correlation between clinical failure and rising in vitro caspofungin MICs was recently observed in patients with C albicans esophagitis (28,29).
Our study has several limitations due to its retrospective nature.The mean duration of five days of fungemia observed in 42% of our candidemic patients may be overestimated because it is possible that clinicians ordered additional blood cultures in sicker patients only.
Mixed fungemia was observed in only three (2%) of our patients.Jensen et al (30) recently reported a 3% rate over a 21-year period, while Al-Rawahi and Roscoe (19) observed a 6% rate over a six-year surveillance period.However, both groups used a differential subculture medium when yeasts were detected in positive blood culture bottles.Such an approach was not used in our study, which could underestimate this phenomenon.The armamentarium of systemic antifungal antibiotics has expanded considerably over the past decade.Our study, as well as observations by others, confirms the high in vitro activity of fluconazole against the large majority of invasive Candida isolates.In patient populations not commonly exposed to azoles, such as medical or surgical unit patients, fluconazole is an appropriate initial therapeutic choice, particularly in clinically stable patients.In unstable patients, for whom organ dysfunction is frequent, and in patient populations in which previous exposure to azoles is common, such as hematology-oncology and HSCT patients, an echinocandin represents a preferred, less toxic alternative to amphotericin B pending the speciation and in vitro susceptibility results.

COnCLuSIOn
No significant changes were observed through this 10-year observational study in the frequency and species distribution of candidemic isolates in our institution.Over the years, C albicans has remained the main cause of fungemia.A tendency toward an increasing number of episodes of Candida BSI in older patients and in patients outside the traditional compromised cancer and critical care patients was noted.In vitro resistance to fluconazole and itraconazole of C albicans remained minimal.The new antifungal antibiotics all have high in vitro activity against our bloodstream Candida isolates.
days of diagnosis occurred in 57 of 185 patients (17 of 58 [29%] between 1996 and 1999; 25 of 68 [37%] between 2000 and 2002; and 15 of 59 [25%] between 2003 and 2006), one of whom was infected with two different Candida species.Death was not significantly associated with age, sex, duration of hospital stay before the episode of candidemia, any of the risk factors listed in Table 2, infecting species (C albicans 34 of 109 [31%]; other species 24 of 81 [30%]) or antifungal resistance.

TabLe 2 Characteristics of patients with candidemia distributed by study period (190 episodes in 185 patients)
*P<0.05; † Combined medical-surgical intensive care units and coronary care unit; ‡ Including the hematopoietic stem cell transplantation (HSCT) unit; § P<0.001

TabLe 4 Risk factors for resistance to fluconazole in 190 episodes of candidemia
*Fluconazole minimum inhibitory concentration 64 mg/L or greater (n=26); † Adjusted for hospital unit at time of diagnosis and previous antifungal use; ‡ P<0.05; § Eleven episodes occurred in outpatients (all infected with fluconazolesusceptible Candida species); medical units include cardiology, endocrinology, gastroenterology, geriatry, internal medicine, nephrology, neurology; surgical units include general surgery and gynecology; ¶ P≤0.001; **Amphotericin B includes deoxycholate (n=13) and lipid-based formulations (n=3).HSCT Hematopoietic stem cell transplantation; NS Not significant