Prospective Study of Candida Species in Patients at a Comprehensive Cancer Center

Abstract

Since most nosocomial systemic yeast infections arise from the endogenous flora of the patient, we prospectively evaluated the species stratification and antifungal susceptibility profile of Candida spp. associated with heavy colonization and systemic infection in patients at Memorial Sloan-Kettering Cancer Center in New York. A total of 349 Candida isolates were obtained from 223 patients during the later half of 1998. Cancer was the most common underlying disease, occurring in 91% of the patients, including 61.8% with organ and 23.7% with hematological malignancies; 4.4% of the patients had AIDS. Candida albicans was the predominant species (67.3%); among 114 non-albicans Candida spp., C. glabrata (45.6%) was the most frequent, followed by C. tropicalis (18.4%), C. parapsilosis (16.6%), and C. krusei (9.6%). The overall resistance to triazole-based agents among all yeast isolates was 9.4 and 10.8% for fluconazole and itraconazole, respectively. A total of 5% of C. albicans strains were resistant to triazole antifungals, whereas 30.8 and 46.2% of C. glabrata strains were resistant to fluconazole (MIC ≥ 64 μg/ml) and itraconazole (MIC ≥ 1 μg/ml), respectively. A significant association was observed between prior treatment with triazole and isolation of fluconazole-resistant C. albicans (P = 0.005, OR 36), although this relationship was not seen in C. glabrata isolates (P = 0.4). This study reinforces the importance of periodic, prospective surveillance of clinical fungal isolates to determine appropriate prophylactic, empiric, and preemptive antifungal therapy for the highly susceptible patient population.

Systemic candidiasis in hospitalized patients has increased steadily over the past four decades and represents a significant cause of morbidity and mortality among severely ill individuals (2, 7, 8, 16, 17, 24, 30). A nationwide increase of 1.8 fungal infections per 1,000 discharges was reported from 1980 to 1990, and 86% were due to Candida spp. (5). Candidemia is the fourth most prevalent cause of bloodstream infections (4, 5, 32), although its attributable mortality (∼40%) exceeds that of bacteremia (19, 32, 40). This rise in fungal infections is exacerbated by the increasing population of immunocommpromised patients, the prevalence of treatment with multiple broad-spectrum antibiotics, and the common use of indwelling intravascular devices (10, 14, 16, 32).

In recent years, this problem has been magnified by an increase in the prevalence of Candida spp. such as Candida glabrata and C. tropicalis, with reduced susceptibility to triazole antifungals, and C. krusei, which is intrinsically resistant to fluconazole and itraconazole (1, 11, 15, 27, 41). In addition, the development of de novo triazole resistance among C. albicans and other normally susceptible species further limits therapeutic options (3, 35, 36). A growing association has been proposed between prior exposure to triazole-based antifungal drugs and development of resistance (1, 22, 42). Recent studies have highlighted important geographic variations in the distribution of Candida species and differences in the prevalence of resistance (26–28).

Given the extensive use of triazoles in hospitals with large populations of cancer patients, it is important to understand changing trends in species distribution and azole susceptibility patterns among Candida spp. In addition, since fungemia due to Candida spp. largely arise from the endogenous flora of the patient, mostly the gastrointestinal and genitourinary tract floras (18, 30, 34, 37, 39), it is important to assess these parameters among the colonizing organisms. We therefore performed a prospective evaluation of Candida associated with prominent colonization and infection in 223 patients at a comprehensive cancer hospital in New York City.

(The results of this study were presented in part at the 9th International Congress on Infectious Diseases, Buenos Aires, Argentina, April 2000.)

Study design.

All clinical isolates that were submitted to the mycology laboratory at Memorial Sloan-Kettering Cancer Center (MSKCC) were screened from 1 July to 31 December 1998. Specimens from sterile body sites and those with prominent colonization (≥50 colonies of Candida) from nonsterile sources were included in the study. More than one isolate from a single patient was included if multiple species or different (genotypic) Candida strains were identified, if the specimens were obtained from separate body sites, or if they resulted from recurrent infection. Demographic information and laboratory data were retrieved from patient charts and from the computerized hospital data system. All specimens were initially processed at the MSKCC Microbiology Laboratory, which isolated and identified Candida spp. Species reidentification (under code) and determination of susceptibility to a panel of antifungal agents were conducted at the New York State Department of Health Mycology Laboratory. All molecular genotyping was carried out at the Public Health Research Institute, New York.

Organism identification.

On identification of yeasts, a germ tube test was performed for presumptive identification of C. albicans. Organisms that failed to form a germ tube were further tested by the auxanographic plate method (BBL Microbiology System, Cockeysville, Md., and Difco Laboratories, Detroit, Mich.). The species not identified by the above methods were subjected to additional standard testing (38). Five to ten colonies were obtained from the primary culture and transported on Trypticase agar slants (Becton Dickinson Microbiology Systems, Cockeysville, Md.) for species reidentification and antifungal susceptibility testing (under code).

Susceptibility testing.

All Candida samples were maintained on Sabouraud dextrose agar media plates (Becton Dickinson Microbiology Systems) at an ambient temperature. A broth microdilution method was performed on the basis of the proposed guidelines of the National Committee for Clinical Laboratory Standards (NCCLS) (20). Antifungal drugs (amphotericin B, flucytosine, ketoconazole, fluconazole, and itraconazole) were obtained from their respective manufacturers. Quality control was performed by testing American Type Culture Collection-designated strains. The interpretative criteria for susceptibility and breakpoints for antifungal drugs were referenced as described by the NCCLS (20).

Statistical analysis.

The association between categorical variables was determined by using Fisher's exact test. A two-sided P value of less than 0.05 was considered statistically significant.

During a 6-month period from 1 July to 31 December 1998, 349 Candida isolates were identified in 329 specimens from 223 patients (Table 1). The median patient age was 60 ± 18.4 years, and the median leukocyte counts were 8.9 ± 8.2/ mm³; there were 110 male and 113 females. Cancer was the most common underlying disease (91%), with organ cancer being present in 61.8% of the patients and hematological malignancies being present in 23.7%. Eighteen patients underwent bone marrow transplantation, while three (14.2%) underwent autologous stem cell reinfusion. Among 10 patients with AIDS, 4 had neoplasms (3 Kaposi's sarcoma and 1 non-Hodgkin's lymphoma). Thirteen patients (5.8%) had benign conditions.

TABLE 1.

Characteristics of patients with Candida colonization or infection

Characteristica	Value
Patients
Total no. of patients	223
No. (%) male	110 (49.3%)
No. (%) female	113 (50.7%)
Age (median ± SD)	60 ± 18.4 yr
Leukocyte count (median ± SD)	8.9 + 8.2/mm3
Underlying diseases
No. (%) with solid-organ cancer	138 (61.8%)
GI tract	38
Lung	30
GU tract	21
Breast	17
Sarcoma	11
GYN	8
CNS	8
Melanoma	3
Endocrine	2
No. (%) with hematological cancer	17 (23.7%)
Lymphoma	17
Acute leukemia	14
Chronic leukemia	11
Multiple myeloma	5
MDS	4
Myelofibrosis	2
No. (%) with AIDS related cancer	10 (4.4%)
No. (%) with benign disease	13 (5.8%)
Specimen source
Total no.	329
Respiratory
Bronchial	80 (24.3%)
B. wash	65
B. lavage	15
Sputum	57 (17.3%)
Oro-pharynx	47 (14.3%)
Pleural fluid	7 (2.1%)
Lung biopsy	3 (0.9%)
Blood	32 (9.7%)
Central venous	16
Peripheral	16
Abdominal and pelvic	62 (18.8%)
Abscess	28 (8.5%)
Stool	12 (3.6%)
Bile	10 (3.0%)
Peritoneal	3 (0.9%)
Wound	8 (2.4%)
Colon biopsy	1 (0.3%)
Urine	23 (6.9%)
Voided	11
Catheterized	12
Integuments	8 (2.4%)
Vagina	5 (1.5%)
CVC line and wound	4
Bone	1 (0.3%)

^aGI, gastrointestinal, GYN, gynecological; CNS, central nervous system; MDS, myelodysplastic syndrome; CVC, central venous catheter. 

C. albicans was the predominant species isolated, accounting for 235 (67.3%) of the 349 Candida isolates (Table 2). Among the 114 non-C. albicans spp., C. glabrata accounted for 45.6%, followed by C. tropicalis (18.4%), C. parapsilosis (16.6%) and C. krusei (3.2%). C. lusitaniae (five isolates), C. lambica (two isolates), C. kefyr (two isolates), C. lipolytica (one isolate), and C. guilliermondi (one isolate) accounted for the remaining 3.2%. Thirty-one Candida spp. were isolated from neutropenic patients; these included C. albicans (38.7%), C. glabrata (25.8%), C. krusei (19.3%), C. parapsilosis (12.9%), and C. lusitaniae (one isolate).

TABLE 2.

Candida susceptibility profile for commonly prescribed antifungal agents

Species (no. [%]) and antifungal	MIC (μg/ml)
	Range	MIC50	MIC90
C. albicans (235 [67.3])
Fluconazole	0.06–>64	0.25	1
Itraconazole	0.03–16	0.03	0.03
Ketoconazole	0.03–16	0.03	0.06
Flucytosine	0.03–64	0.25	2
Amphotericin B	0.03–0.5	0.25	0.5
C. glabrata (52 [14.8])
Fluconazole	0.5–>64	16	>64
Itraconazole	0.03–>16	0.5	4
Ketoconazole	0.03–16	0.5	2
Flucytosine	0.03–1	0.12	0.25
Amphotericin B	0.03–2	0.25	0.5
C. tropicalis (21 [6.0])
Fluconazole	0.12–64	1	64
Itraconazole	0.03–16	0.06	16
Ketoconazole	0.03–16	0.03	2
Flucytosine	0.06–64	0.25	0.5
Amphotericin B	0.03–1	0.25	0.5
C. krusei (11 [3.1])
Fluconazole	8–>64	32	>64
Itraconazole	0.12–2	0.25	0.5
Ketoconazole	0.25–1	0.25	0.5
Flucytosine	0.06–1	8	16
Amphotericin B	0.12–2	0.5	1
Othersa (30 [8.5])
Fluconazole	0.03–32	1	16
Itraconazole	0.03–0.25	0.03	0.25
Ketoconazole	0.03–0.5	0.03	0.12
Flucytosine	0.03–32	0.06	16
Amphotericin B	0.03–0.5	0.12	0.25

^aOthers include 19 C. parapsilosis isolates, 5 C. lusitaniae isolates, 2 C. lambica isolates, 2 C. kefyr isolates, one C. lipolytica isolate, and one C. guilliermondii isolate. 

Candida was most frequently isolated from upper respiratory tract specimens (104 specimens, including 57 sputum and 47 oropharyngeal samples). A total of 195 bronchial tree specimens were cultured during the study period, and 80 (41%) were positive for Candida (Table 1). Yeast was also isolated in 5% of bile samples (10 of 199) and 4.6% of vaginal cultures (5 of 109). In fungal stool cultures, 83.3% yielded yeast with either a paucity or absence of normal bacterial flora. Nearly half (52.1%) of urine cultures were from samples from catheterized patients; they revealed seven C. albicans, three C. glabrata, two C. tropicalis, and one C. parapsilosis isolates.

During the course of this study, the incidence of nosocomial candidemia was 3.3% (32 of 958) among all episodes of bloodstream infections. Of 18 fungemic patients, 83% had intact neutrophil counts (leukocytes > 1.0/mm³); 12 of these had an underlying solid organ cancer, 4 were diagnosed with leukemia, and 1 each had non-Hodgkin's lymphoma and myelodysplastic syndrome. Half of the blood specimens (16 specimens) were drawn from a central venous catheter. C. albicans was the predominant species (43.8%), followed by C. parapsilosis (18.8%), C. lusitaniae (12.5%), C. glabrata (9.3%), C. tropicalis (9.3%), and C. krusei (6.3%).

The drug susceptibility data for the clinical isolates are summarized in Table 2. Among the 349 Candida isolates tested, all except 2 (C. glabrata and C. krusei) were susceptible to amphotericin B (MIC ≤ 0.5 μg/ml). The overall triazole resistance to fluconazole and itraconazole was 9.4 and 10.8%, respectively. A total of 95% of C. albicans isolates were susceptible to fluconazole (MIC ≤ 8 μg/ml) and itraconazole (MIC ≤ 0.25 μg/ml), while resistance was 3.4% to fluconazole (MIC ≥ 64 μg/ml) and 4% to itraconazole (MIC ≥ 1 μg/ml). Eight isolates (3.4%) were resistant to flucytosine, and 2% showed in vitro resistance to ketoconazole.

Resistance among C. glabrata isolates to fluconazole and itraconazole was 30.7 and 46.2%, respectively, whereas 19% of C. tropicalis isolates were resistant to fluconazole and 21% were resistant to itraconazole. All antifungal agents were effective against C. parapsilosis. The MIC for 90% of isolates (MIC₉₀) of fluconazole ranged from 1 μg/ml for C. albicans to ≥64 μg/ml for C. glabrata, C. tropicalis, and C. krusei (Table 2). The itraconazole MIC₉₀ was higher for C. tropicalis (16 μg/ml) C. glabrata (4 μg/ml), and C. krusei (0.5 μg/ml). The antifungal susceptibility profile among the bloodstream infection isolates was similar to that of the total isolate population reported in Table 2.

Of 22 C. glabrata isolates for which the fluconazole MIC was ≥32 μg/ml, 10 (45.4%) were isolated from patients with prior exposure to triazole-based antifungals for a duration of >7 days in the preceding 12 weeks. On the other hand, only 7 of 33 fluconazole-susceptible (≤16 μg/ml) C. glabrata isolates (21.8%) were obtained from triazole-experienced patients (P = 0.4). For fluconazole-resistant C. albicans isolates, 85.7% were obtained from patients with prior triazole exposure (P < 0.005) (Table 3). Molecular genotyping (by random amplification of polymorphic DNA) confirmed that the resistant isolates of C. albicans and C. glabrata were distinct strains (data not shown), indicating that they were not transmitted from common progenitor strains.

TABLE 3.

Association between prior exposure to triazole-based antifungals and isolation of fluconazole resistant Candida strains

Species	No. of patients with isolates for which fluconazole MIC (μg/ml) wasa:		P	Odds ratio
	≥32	≤16
C. glabrata	10/22 (45.4%)	7/33 (21.8%)	>0.4	NCb
C. albicans	6/7 (85.7%)	2/10 (20.0%)	0.005	36

^aThe numerator indicates the number of patients with prior exposure to fluconazole, and the denominator indicates the total number of patients in the group. 

^bNC, noncalculable. 

Nosocomial fungal infections adversely affect the outcome of underlying disease and significantly increase the cost of care of hospitalized patients (31, 39, 40, 45). Agranulocytosis, indwelling central venous catheters, extensive ulceration of mucous membranes, and treatment with multiple broad-spectrum antibiotics are important contributing factors in this setting (10, 14, 16, 33). In addition, Candida colonization in most instances precedes fungemia and is regarded as an independent risk factor for systemic fungal infection (12, 30, 37, 39, 40).

With the introduction of triazole therapy in the early 1990s, there has been a declining trend in the proportion of C. albicans among total clinical Candida isolates, accounting for just over 50% in most studies (5, 21, 24, 25, 28–30). In this study, C. albicans was the dominant organism, representing 67% of all Candida spp. isolated (Table 2). Among the non-albicans Candida spp. in our survey, nearly half (46%) were C. glabrata, followed by C. tropicalis (18.4%) and C. parapsilosis (16.6%) (Table 2). This distribution is similar to the results of the recent Surveillance and Control of Pathogens of Epidemiologic Importance (SCOPE surveillance program), involving 50 medical centers throughout the United States over a 14-month period during 1995 and 1996 (31, 32). The emergence of C. glabrata as the principal non-C. albicans sp. in our patients was not surprising, given its ability to develop resistance to azole-based drugs (11, 27). In fact, we report that for more than 60% of the isolates, the fluconazole MIC was ≥16 μg/ml and that 83% showed reduced susceptibility to itraconazole (Table 2). Recently, an increase in the prevalence of C. krusei infections has been reported from several institutions (6, 13, 24, 41), although the prevalence of this species remained <5% among all clinical Candida isolates at MSKCC (Table 2).

The level of triazole resistance in C. albicans (3% fluconazole resistance and 4% itraconazole resistance) at our institution compares favorably with 5.5% fluconazole resistance reported from institutions in the northeastern United States during 1995 to 1996 (25). The susceptibility profiles of C. parapsilosis and C. tropicalis were consistent with findings from other institutions (21, 27). Interestingly, all C. lusitaniae (5) isolates were susceptible to amphotericin B (MIC₉₀, 0.25 μg/ml), a finding inconsistent with other reports (41, 42).

There is evidence linking prior exposure to triazoles and subsequent emergence of drug resistance in Candida spp. (1, 42). It was expected that such an association would be found for C. glabrata, given the high prevalence of triazole resistance in this study. Nonetheless, we report no significant association between prior fluconazole or itraconazole prophylaxis or treatment and high-level resistance among C. glabrata strains (P = 0.4; Table 3). The fact that resistance among C. glabrata isolates was observed independently of triazole exposure may reflect a selection of resistant isolates from the environmental pool, most probably within the community. Alternatively, it may relate to a rapid acquisition of resistance traits in colonizing strains due to the haploid nature of the organism and its ability to mutate rapidly (9). In contrast, C. albicans isolates displaying resistance (<5%) to fluconazole and itraconazole were isolated predominantly from patients with prior triazole exposure (P = 0.005) (Table 3).

These results emphasize the dynamic nature of Candida spp. in the compromised host and the complexity of factors promoting drug resistance. Furthermore, they emphasize the need for ongoing surveillance of antifungal susceptibility patterns and an evaluation of factors influencing the introduction of resistance among clinically significant yeasts in highly susceptible patients. Interval surveillance of this type is an essential component in developing institutional guidelines for prophylaxis and empiric or preemptive therapy for these life-threatening infections.