JOURNAL OF CLINICAL MICROBIOLOGY, Feb. 2003, p. 735–741 0095-1137/03/$08.00⫹0 DOI: 10.1128/JCM.41.2.735–741.2003 Copyright © 2003, American Society for Microbiology. All Rights Reserved.

Vol. 41, No. 2

Candida tropicalis in a Neonatal Intensive Care Unit: Epidemiologic and Molecular Analysis of an Outbreak of Infection with an Uncommon Neonatal Pathogen Emmanuel Roilides,1,2 Evangelia Farmaki,1 Joanna Evdoridou,3 Andrea Francesconi,2 Miki Kasai,2 Joanna Filioti,1 Maria Tsivitanidou,4 Danai Sofianou,4 George Kremenopoulos,3 and Thomas J. Walsh2* Third Department of Pediatrics1 and First Neonatology Department,3 Aristotle University of Thessaloniki, and Microbiology Department, Hippokration Hospital,4 54642 Thessaloniki, Greece, and Pediatric Oncology Branch, National Cancer Institute, National Institutes of Health, Bethesda, Maryland 208922 Received 14 August 2002/Returned for modification 11 September 2002/Accepted 21 November 2002

From June to July 1998, two episodes of Candida tropicalis fungemia occurred in the Aristotle University neonatal intensive care unit (ICU). To investigate this uncommon event, a prospective study of fungal colonization and infection was conducted. From December 1998 to December 1999, surveillance cultures of the oral cavities and perinea of the 593 of the 781 neonates admitted to the neonatal ICU who were expected to stay for >7 days were performed. Potential environmental reservoirs and possible risk factors for acquisition of C. tropicalis were searched for. Molecular epidemiologic studies by two methods of restriction fragment length polymorphism analysis and two methods of random amplified polymorphic DNA analysis were performed. Seventy-two neonates were colonized by yeasts (12.1%), of which 30 were colonized by Candida albicans, 17 were colonized by C. tropicalis, and 5 were colonized by Candida parapsilosis. From December 1998 to December 1999, 10 cases of fungemia occurred; 6 were due to C. parapsilosis, 2 were due to C. tropicalis, 1 was due to Candida glabrata, and 1 was due to Trichosporon asahii (12.8/1,000 admissions). Fungemia occurred more frequently in colonized than in noncolonized neonates (P < 0.0001). Genetic analysis of 11 colonization isolates and the two late blood isolates of C. tropicalis demonstrated two genotypes. One blood isolate and nine colonization isolates belonged to a single type. The fungemia/colonization ratio of C. parapsilosis (3/5) was greater than that of C. tropicalis (2/17, P ⴝ 0.05), other non-C. albicans Candida spp. (1/11, P ⴝ 0.02), or C. albicans (0/27, P ⴝ 0.05). Extensive environmental cultures revealed no common source of C. tropicalis or C. parapsilosis. There was neither prophylactic use of azoles nor other risk factors found for acquisition of C. tropicalis except for total parenteral nutrition. A substantial risk of colonization by non-C. albicans Candida spp. in the neonatal ICU may lead to a preponderance of C. tropicalis as a significant cause of neonatal fungemia. Candida spp. are common causes of nosocomial invasive infections in neonatal intensive care units (ICUs). Although Candida albicans has historically been the most frequently isolated species, infections caused by non-C. albicans Candida spp. have been diagnosed with increased frequency in recent years (14, 17). In particular, Candida parapsilosis has become the predominant fungal pathogen in many neonatal ICUs (11, 14, 17). Moreover, recent studies have shown that C. parapsilosis is often a cause of clusters and common-source outbreaks (12, 29). In contrast, little is known about the roles of other non-C. albicans Candida spp. in neonatal ICUs (4). To our knowledge, Candida tropicalis fungemia in neonates has not been adequately described, and the significance of mucocutaneous colonization of neonates by this pathogen remains speculative. Two chronologically related unusual cases of C. tropicalis fungemia that occurred in the Aristotle University neonatal ICU prompted us to initiate a prospective study of fungal colonization and infection in the ICU. In particular, we aimed

to investigate this new trend of candidemia due to non-C. albicans Candida spp. and to examine whether mucocutaneous colonization correlates with the apparent increase in the isolation of C. tropicalis from bloodstream infections. MATERIALS AND METHODS Epidemiological studies. After the occurrence of two chronologically related cases of C. tropicalis fungemia in the neonatal ICU (Table 1, patients 1 and 2), a prospective study of fungal colonization and infection was initiated. The study was undertaken after approval was obtained from the Ethics Committee of Hippokration Hospital. Between December 1998 and December 1999, a total of 781 neonates were admitted to the neonatal ICU. Of these, the 593 who were expected to stay ⬎1 week in the neonatal ICU were enrolled in the study. On admission and at weekly intervals thereafter, 3,194 samples were taken from the oral cavity, perineum, and tracheal aspirates. Oral and perineal samples were obtained by using premoistened cotton-tipped swabs and placed in Culturette transport medium (Euromed, Letse, Italy). Tracheal aspirates were collected in sterile traps during airway suctions. All specimens were then cultured on Sabouraud glucose agar (Scharlau Chemie S.A., Barcelona, Spain) with 50 mg of chloramphenicol per liter and 50 mg of gentamicin per liter at 37°C for 48 h. Potential environmental reservoirs such as nearby surfaces and equipment, including benches, sinks, ventilators, cubicles, and intravenous fluid administration pumps, were cultured by using premoistened cotton-tipped swabs. Both hands of staff members, including neonatal ICU physicians and nurses, physicians rotating through the unit, and ancillary personnel of the unit, were also cultured for fungi by the same technique at least twice during the surveillance study period. A sterile cotton swab was moistened with sterile normal saline, and

* Corresponding author. Mailing address: Immunocompromised Host Section, Pediatric Oncology Branch, National Cancer Institute, Bldg. 10, Rm. 13N240, Bethesda, MD 20892. Phone: (301) 402-0023. Fax: (301) 402-0575. E-mail: [email protected]. 735

736

ROILIDES ET AL.

J. CLIN. MICROBIOL. TABLE 1. Cases of neonatal fungemia due to C. tropicalisa

Patient

Sex

BW (g)

GA (wk)

1

M

3,110

36

2

M

1,490

3 4

M M

1,290 810

a

Onset of infection (day/mo/yr)

Age at onset (days)

No of positive cultures

3/6/98

12

1

31

12/6/98

20

1

28 29

17/12/98 18/3/99

12 14

2 2

Treatment (duration [days])

Outcome

AmB, 1 mg/kg (NA)

Cure

AmB, 1 mg/kg (10)

Cure

AmB, 1 mg/kg (34) AmB, 1 mg/kg (14)

Cure Cure

Risk factors

Asphyxia, IPPV, pleural tube, CVC, TPN, prior antibiotic use IPPV, CVC, TPN, Escherichia coli sepsis IPPV, CVC, TPN, Klebsiella sepsis IPPV, CVC, TPN, Klebsiella sepsis

Abbreviations: M, male; TPN, total parenteral nutrition; AmB, deoxycholate amphotericin B; NA, not available.

the entire surface of the hand, including the area under the fingernails, was swabbed. A separate swab was used for each hand. Possible risk factors for the acquisition of C. tropicalis as colonizers or as infectious pathogens were searched for in the clinical charts of the patients. Briefly, demographic data, device days for ventilators and central venous catheters (CVCs), use of antibiotics, total parenteral nutrition (NADP), corticosteroids, and other medications were recorded. Mycological studies. Fungal isolates from colonization cultures were identified to the species level by the ability to form germ tubes and by one or more of the following methods of Candida sp. identification: the Chromagar Candida test (Becton Dickinson, Sparks, Md.), the Mycotube test (Roche, Basel Switzerland), and the API ID 32C test (bioMe´rieux SA, Marcy l’Etoile, France). Fungal isolates from blood cultures were identified to the species level by use of the germ tube test and the API ID 32C test. To test the susceptibility of isolates to amphotericin B, flucytosine, fluconazole, and itraconazole, the M-27A broth microdilution method for MIC determination standardized by the NCCLS was used with appropriate control strains (22). In addition, minimal fungicidal concentrations were also determined. Genetic studies. Molecular epidemiological studies were performed by application of four different methods, i.e., two methods of restriction fragment length polymorphism (RFLP) of genomic DNA digested with restriction enzyme HindIII or BstNI and two PCR methods using either short tandemly repeated sequences [(GACA)4] or a random primer (RP02). Control C. tropicalis isolates ATCC 750 and ATCC 1369 were obtained from the American Type Culture Collection (Vienna, Va.), and isolate NIH 719 (NIH 03180719) was obtained from the Clinical Microbiology Laboratory at the Warren Grant Magnuson Clinical Center, National Institutes of Health (Bethesda, Md.). DNA extraction. Flasks containing 2% Sabouraud glucose broth were inoculated with two or three colonies and incubated in a 30°C shaking water bath overnight. Cells were harvested, and spheroplasts were prepared by resuspending the cells in 4 ml of spheroplast buffer containing 0.25 mg of Lyticase (Sigma L-8137) per ml as previously described (18). DNA was extracted from spheroplasts (100 mg [wet weight]) by using the Qiagen DNeasy Plant mini DNA extraction kit (Qiagen, Santa Clarita, Calif.). The manufacturer’s protocol was followed, with the following exceptions. (i) In step 2, 0.2 g of glass beads (G-8772; Sigma) was added to the sample and vortexed at high speed for 10 min. (ii) In step 4, samples were centrifuged for 5 min at high speed before application of the lysate to the QIAshredder spin column. (iii) In step 12, DNA was eluted twice from each column by using 35 ␮l of AE elution buffer. RFLP. On the basis of the yield of DNA (visualized on a 1% agarose gel), restriction digests were set up in accordance with the manufacturer’s directions by using either 10 U of HindIII or BstNI restriction enzyme. Samples digested with HindIII were incubated at 37°C for 4 h, and BstNI samples were incubated at 60°C for 4 h. All digested samples were run on 0.7% agarose gels (50% I.D.NA agarose [catalog no. 50170; FMC Bioproducts] and 50% ultra PURE agarose [catalog no. 15510-027; Gibco BRL]) in 1⫻ Tris-borate-EDTA (TBE). Gels were run for 22 h at 45 V. Agarose gels were stained in fresh 1⫻ TBE containing 0.4 ␮g of ethidium bromide per ml and destained as needed. Random amplified polymorphic DNA. Two different oligonucleotide primers were used in separate reactions, RP02 (5⬘ GCG ATC CCC A 3⬘) and (GACA)4 (30). Amplification reactions (50 ␮l) were performed in accordance with the manufacturer’s protocol, with the following exceptions: 2.5 U of Taq DNA polymerase (Boehringer Mannheim, Indianapolis, Ind.), 0.1 mg of bovine serum albumin per ml, and 1 ␮l of genomic DNA diluted 1:10 in Tris-EDTA. Samples were incubated at 37°C for 15 min. PCR was performed for the RP02 primer as follows: 94°C for 5 min; 36°C for 5 min and 72°C for 5 min for 4 cycles; 94°C for 1 min., 36°C for 1 min, and 72°C for 2 min for 30 cycles; followed by a 10-min extension period at 72°C; ending in a 4°C soak. PCR was performed for the

(GACA)4 primer as follows: 93°C for 2 min; 93°C for 0.3 min, 40°C for 1 min, and 72°C for 0.5 min for 40 cycles; followed by a 6-min extension period at 72°C; ending in a 4°C soak. Both random amplified polymorphic DNA reactions (15 ␮l) were run on a 2% agarose gel containing 1⫻ TBE plus ethidium bromide (0.4 ␮g/ml). Molecular epidemiological analysis. Analysis of gel banding patterns was performed by visualization of images captured by an AlphaImager (Alpha Innotech Corp., San Leandro, Calif.). The images were captured as either positive or negative for the best resolution of banding patterns. The banding patterns on gels from DNA of each isolate were compared for relatedness to all other isolates. Investigators who were blinded to the identity of isolates performed the comparison of gels. Comparative banding patterns were assessed independently for each method (each method was performed a minimum of three times) and compiled into a consensus biotyping pattern for each isolate. Molecular relatedness was defined as an identical consensus biotyping pattern. Statistical analysis. The statistical program GraphPad Instat (Graphpad Inc., San Diego, Calif.) was used for analysis. Statistical evaluation of differences in proportions and odds ratio (OR) and 95% confidence interval (CI) calculations were performed by Fisher’s exact test. Differences in means were evaluated by means of the Student t test. A P value of ⱕ0.05 indicated statistical significance. All of the P values reported are two sided.

RESULTS Fungal colonization. A total of 3,194 colonization cultures were processed during the study period of December 1998 to December 1999. Of these, 1,536 were obtained from the oral cavities and 1,536 were from the perinea of all 593 neonates. An additional 122 tracheal aspirates from a subgroup of 74 neonates were cultured. There were 317 (53.5%) male and 276 (46.5%) female neonates. The mean birth weight (BW) was 2,155 ⫾ 35.5 g, and the mean gestational age (GA) was 34 ⫾ 0.2 weeks. During the study period, colonization by yeasts was detected in 72 of the 593 neonates, yielding an overall colonization rate of 12.1%. C. albicans was isolated from 30 neonates (42% of the colonized neonates), C. tropicalis was isolated from 17 (24%), Candida krusei was isolated from 8 (11%), C. parapsilosis was isolated from 5 (7%), and Candida glabrata was isolated from 5 (7%). Three neonates were colonized by two yeasts each: C. albicans and C. glabrata colonized one, C. albicans and Trichosporon sp. colonized one, and C. tropicalis and C. lusitaniae colonized one. four neonates were colonized by other yeasts; specifically, Trichosporon sp. colonized two, Candida guilliermondii colonized one, and an unidentified, germ tube-negative yeast colonized one. Fungal infections. During the study period (December 1998 to December 1999), 781 neonates were admitted to the neonatal ICU. Of these, 10 had an equal number of fungemia episodes, yielding an incidence rate of 12.8/1,000 admissions. The most common species isolated from these episodes was C. parapsilosis (six cases), followed by C. tropicalis (two cases), C.

C. TROPICALIS IN A NEONATAL ICU

VOL. 41, 2003 TABLE 2. Characteristics of C. tropicalis-colonized neonates Characteristic

No. of colonized neonates.....................................................

8 (47)a 9 (53)

No. of vaginal deliveries ........................................................ No. of cesarean sections........................................................

9 (53) 8 (47)

No. of colonization isolates................................................... No. of oral cavity isolates...................................................... No. of perineum isolates .......................................................

52 29 23

Time of colonization after admission Early (⬍7 days) .................................................................. Late (⬎7 days)....................................................................

2 15

Values in parentheses are percentages.

glabrata, and Trichosporon asahii (one case each). No episodes due to C. albicans occurred (Fig. 2). The characteristics of the four neonates with fungemia due to C. tropicalis are shown in Table 1. Invasive fungal infections occurred in previously colonized neonates more frequently than in noncolonized neonates. Thus, 7 of 10 neonates with a fungal infection were also colonized by yeasts. On the other hand, 3 out of 521 noncolonized neonates (0.6%) suffered from infection, compared to 7 out of 72 colonized neonates (9.7%; OR ⫽ 0.054; 95% CI, 0.013 to 0.213; P ⬍ 0.0001). Characteristics of neonates colonized by C. tropicalis, invasiveness of Candida spp., and antifungal drug susceptibility of C. tropicalis isolates. The neonates colonized by C. tropicalis tended to have lower BWs and GAs than the neonates colonized by other yeasts (Table 2). However, these differences were not significant. Only 2 (12%) of 17 neonates were colonized by C. tropicalis at an early stage (⬍7 days). The remaining neonates were colonized at a late stage (⬎7 days) of their neonatal ICU stay. As measured by the fungemia/colonization ratio, there were significant differences in invasiveness among C. albicans, C. tropicalis, and C. parapsilosis, with C. parapsilosis being the most invasive and C. albicans being the least invasive species. In particular, of 5 neonates colonized by C. parapsilosis, 3 progressed to fungemia, whereas of 17 neonates colonized by C. tropicalis, only 2 progressed to fungemia. This fungemia/colonization ratio of C. parapsilosis was significantly greater than that of C. tropicalis, other non-C. albicans Candida spp., or C. albicans (Table 3). During the first 4 months of the study, C. tropicalis caused a cluster of two cases of fungemia (Table 1, patients 3 and 4) and 17 cases of colonization (52 isolates from multiple sites and times of colonization). According to data obtained by reviewing the laboratory records, the four cases included in Table 1 were the only cases of candidemia due to C. tropicalis diagnosed in the neonatal ICU. No other episodes of candidemia due to this Candida species were diagnosed during the 5.5 years prior to the cases described and during the subsequent months.

No. of No. of Fungemia/ colonized P colonized colonization and fungemic valuea neonates ratio neonates

Candida sp.

17

No. of males............................................................................ No. of females ........................................................................

a

TABLE 3. Progression from colonization to fungemia

Value

Mean BW ⫾ SEM (g) ........................................................... 1,637 ⫾ 247 Mean GA ⫾ SEM (wk)......................................................... 30.4 ⫾ 2.7

737

C. parapsilosis C. tropicalis C. albicans Other non-C. albicans spp.

5 17 27 11

3 2 0 1

0.60 0.12 0 0.09

0.05 0.05 0.02

a Difference from C. parapsilosis (Fischer’s exact test). Six other neonates were colonized by more than one species, and none of them progressed to fungemia.

All cases of C. tropicalis fungemia were successfully treated with amphotericin B (Table 1). It is noteworthy that of 3 blood isolates and 19 colonization isolates studied for susceptibility to antifungal agents, all were susceptible to amphotericin B and flucytosine. With regard to azoles, one blood and five colonization isolates were nonsusceptible to fluconazole and the same blood isolate and 15 colonization isolates were nonsusceptible to itraconazole. Specifically, four and six isolates exhibited dose-dependent susceptibility to fluconazole and itraconazole, respectively, whereas one and nine isolates were resistant to the two azoles, respectively (Table 4). Genetic analysis of C. tropicalis isolates. Genetic analysis of two blood isolates and 12 randomly selected colonization isolates derived from six neonates demonstrated two genotypes, A and B (Fig. 1). One blood and all colonization isolates derived from five of six colonized neonates belonged to genotype B. Only one patient was colonized and infected by genotype A (Table 5, patient 1). Environmental surveillance. Extensive environmental surveillance, including cultures from personnel hands, benches, sinks, ventilators, cubicles, and intravenous fluid administration pumps, revealed no common source of C. tropicalis or C. parapsilosis. Antifungal azoles were not used in the neonatal ICU prophylactically, nor was there found a significant role of other possible risk factors for their acquisition, such as central CVCs, intermittent positive pressure ventilation (IPPV), prematurity, prior sepsis, and administration of antibiotics. Only prior administration of NADP was found to be a possible risk factor for C. tropicalis colonization. Specifically, of 17 neonates colonized by C. tropicalis, 6 (35%) had received NADP (mean duration of NADP administration to 17 neonates, 2.18 days), compared to only 1 (3.7%) out of 27 neonates colonized by C. albicans (mean duration of NADP administration to 27 neonates, 0.19 day). The difference between the two percentages is significant (OR ⫽ 14.2; 95% CI, 1.5 to 132.1; P ⫽ 0.009). TABLE 4. Susceptibility of C. tropicalis to antifungal agentsa Drug

50%

Amphotericin B 0.25 Flucytosine ⬍0.25 Fluconazole 2 Itraconazole 0.25 a b c

MFCb (␮g/ml)

MIC (␮g/ml) c

90%

c

0.5 0.5 24 8

Range

50%

90%

Range

0.125–0.5 0.5 1 0.25–1 ⬍0.25–2 1 16 0.25–16 0.25–⬎128 128 ⬎128 4–⬎128 0.06–⬎32 ⬎32 ⬎32 32–⬎32

Three blood and 19 colonization isolates of C. tropicalis were studied. MFC, minimal fungicidal concentration. Percentage of tested isolates inhibited or killed by the stated concentration.

738

ROILIDES ET AL.

J. CLIN. MICROBIOL.

FIG. 1. Molecular analysis of C. tropicalis colonization and fungemia cluster by RFLP and PCR. (A) Analysis of isolates by RFLP of genomic DNA digested with HindIII. (B) Analysis of isolates by RFLP of genomic DNA digested with BstNI. (C) Analysis of isolates by PCR using short tandemly repeated sequences (GACA)4. (D) Analysis of isolates by PCR using a random primer, RP02. Panels A and C are negative images for best resolution of banding patterns. For the same reason, panels B and D are positive images. Markers, lanes 1 and 19; patient 1 (Pt1), lanes 2, 3, and 4; patient 2, lanes 5 and 6; patient 3, lanes 7 and 8; patient 4, lane 9; patient 5* (C. albicans), lane 10; patient 6, lanes 11 and 12; patient 7, lanes 13, 14, and 15; laboratory control isolates of C. tropicalis (NIH 719, ATCC 750, and ATCC 1369), lanes 16, 17, and 18. Table 5 shows the source of each strain depicted as a three-digit number above the lanes.

DISCUSSION In this report, we have described a cluster of colonization and fungemia due to C. tropicalis in a single neonatal ICU and an analysis of the epidemiology of these events. We found that the invasiveness of C. tropicalis was moderate and that the outcome of the infection was fair. Acquisition of C. tropicalis very likely occurred in the neonatal ICU by cross-contamination. In contrast to C. parapsilosis, which can cause clusters and common-source outbreaks (3, 12, 19, 33), the potential of C. tropicalis for nosocomial transmission has not been well studied. During the 4-month early period of our prospective study, we encountered a cluster of two cases of fungemia and 17 cases of colonization. Two more cases had occurred 6 months earlier (Table 1). To our knowledge, this is the second cluster of C. tropicalis fungemia of neonates described in the English literature (6). However, this is the largest cluster involving 4 neonates with fungemia and 17 neonates colonized by C. tropicalis.

Over the past years, there has been a change in the distribution of Candida spp. that have caused nosocomial infections in the neonatal ICU setting. Although C. albicans remains the most frequently isolated Candida species in several centers, the role of other Candida species is increasing (14, 17). In many neonatal ICUs, C. parapsilosis has emerged as the predominant pathogen causing candidemia in neonates (11, 14, 17, 24). Moreover, there are several reports of nosocomial cross-infections due to C. albicans or C. parapsilosis in the neonatal ICU setting (5, 10, 19, 25, 26, 33). In contrast, C. tropicalis is rarely encountered in neonates, unlike adults and children with hematological malignancies, in whom it causes infections with a high mortality rate (7, 16, 32, 34). The epidemiology of C. tropicalis in neonates is unclear, but the potential for nosocomial transmission must be considered. In our neonatal ICU, the incidence of fungemia was 12.8 cases per 1,000 admissions. Historically, C. albicans has been

C. TROPICALIS IN A NEONATAL ICU

VOL. 41, 2003 TABLE 5. Molecular genetic analysis of C. tropicalis fungemia and colonization in a neonatal ICU Patient and isolate no.a

1

2

3

4 5 6

7

Source

Biotyping pattern Isolation date (day/mo/yr) HindIII BstNI (GACA) RP02 4

498 503 663

Perineum Oral cavity Blood

12/12/98 12/12/98 17/12/98

A A A

A A A

A A A

A A A

517 523

Oral cavity Oral cavity

5/2/99 16/2/99

B B

B B

B B

B B

519 581

Oral cavity Perineum

15/2/99 15/3/99

B B

B B

B B

B B

587

Oral cavity

17/3/99

B

B

B

B

555b

Perineum

9/3/99

C

C

C

C

529 577

Oral cavity Oral cavity

23/2/99 15/3/99

B B

B B

B B

B B

583 585 661

Perineum Oral cavity Blood

17/3/99 17/3/99 18/3/99

B B B

B B B

B B B

B B B

18/3/99

D E F

D E F

D E F

D E F

Controls NIH 719c Lung ATCC 750c ATCC 1369c a

Patients 1 and 7 are patients 3 and 4 in Table 1. C. albicans blinded control strain. c C. tropicalis control strains. b

the most common species recovered, followed by C. parapsilosis (27). However, the isolation of non-C. albicans Candida spp. as causes of fungemia has increased during the past 4 years. During the study period, C. parapsilosis and C. tropicalis have emerged as the predominant species, accounting for 80% of infected neonates. Notably, no episodes due to C. albicans have occurred in the same period. There are several possible explanations for this pathogen shift away from C. albicans toward other species, including several diagnostic and therapeutic interventions, such as the increasing use of CVCs and NADP. For example, C. parapsilosis has been significantly associated with the use of NADP and the presence of CVCs (3, 12). In addition, the extensive use of antifungal prophylaxis may have played an important role in this changing pattern of fungal pathogens isolated in some centers (8). However, there was no antifungal drug prophylaxis practiced in our neonatal ICU, which may have promoted a predominance of less susceptible non-C. albicans Candida spp. In addition, no significant role of other risk factors for C. tropicalis acquisition, except prior administration of NADP, was found. Recent data have shown that C. tropicalis is the second or third leading cause of candidemia in adults, especially in patients with lymphoma, leukemia, and diabetes. In a prospective, population-based surveillance for candidemia in two

739

United States cities, C. tropicalis infection was most often seen in persons with cancer and diabetes mellitus (46 and 24% of cases of candidemia due to C. tropicalis) (13). In a retrospective analysis of 81 episodes of candidemia in children that occurred over a 7-year period (17), C. tropicalis was the cause of candidemia in 10% of the episodes. In this study, C. tropicalis was more likely to be isolated from older children than from neonates (17% of children versus 3% of neonates) but this difference was not significant. In one study, C. tropicalis accounted for 20% of the episodes of neonatal fungemia (20) but no further analysis of these cases was presented. While two other studies reported C. tropicalis in neonates (6, 21), this study describes the largest number of C. tropicalis isolates and is the only study that included comprehensive genotyping of these isolates. Two large prospective multicenter studies have shown that while C. tropicalis is the second or third leading cause of candidemia in adults, it is infrequent among neonates (2, 24) and this difference was significant (19% in adults versus 0% in neonates, P ⫽ 0.04) (24). C. tropicalis has a propensity to produce fungemia in neutropenic patients and appears to cause a disproportionate number of serious infections in patients with hematological malignancies (34). Most infections appear to originate from the patients’ alimentary tract microflora (32). The higher proportion of C. tropicalis in adult patients and older children hospitalized in oncology-hematology units might be partially explained by the prophylactic or empirical use of amphotericin B (23). However, other differences, such as host defense deficiency and mucosal integrity, may be related to this discrepancy. It is noteworthy that while all of the neonatal C. tropicalis isolates tested were susceptible to amphotericin B and flucytosine, a significant portion of them was found to possess reduced susceptibility to azoles, especially itraconazole, compared to C. albicans. C. tropicalis appears to be more virulent than C. albicans in patients with hematological malignancies, and disseminated infection is associated with high mortality rates (15, 16, 31, 32). Among adults with or without cancer, systemic infections due to C. tropicalis have been associated with higher rates of mortality and dissemination than infection due to C. parapsilosis (2, 9). Indeed, the virulence of the latter is limited compared with that of C. albicans and C. tropicalis, and thus, the mortality caused by C. parapsilosis is lower than that caused by other Candida spp. It is noteworthy that all four neonates with C. tropicalis fungemia survived after antifungal treatment with amphotericin B. Once introduced into a host with an impaired immune system, C. tropicalis may be more virulent than C. albicans and proceed from colonization to invasion more easily. A verylow-BW (VLBW) infant with an immature immune system may behave like an immunocompromised adult patient in this regard (34). Several studies have shown that rates of fungal colonization among VLBW neonates vary greatly, with C. albicans and C. parapsilosis being the most prevalent species (1, 10, 28). In a prospective study of fungal colonization in VLBW neonates, C. tropicalis was detected in 7.7% of colonized neonates (1), although others have found a lower percentage (28). We found a relatively high incidence of C. tropicalis colonization (24%). The high prevalence in our study may be due to an outbreak or

740

ROILIDES ET AL.

may be associated with subtle changes in the diagnostic and treatment approaches used in the unit; however, we did not find any risk factors, other than previous use of NADP, that particularly characterized neonates with C. tropicalis. Patient-staff interaction has been proposed on several occasions as a mechanism of cross-infection. This pattern of acquisition suggests transmission to infants once they are within the unit. In our study, only 2 (12%) out of 17 neonates were colonized by C. tropicalis at an early stage, possibly through vertical transmission; the remaining neonates were colonized at a late stage (⬎7 days) of their neonatal ICU stay, suggesting nosocomial acquisition of C. tropicalis. Although the examination of the hands of health care workers in our study failed to document C. tropicalis colonization, transient hand colonization of personnel cannot be excluded and, indeed, may have played an important role in this cluster of colonization in the neonatal ICU. Nevertheless, the importance of hand washing and compliance with guidelines for preventing nosocomial infections was emphasized to the personnel at the time the cluster was investigated, possibly causing the personnel to take stricter precautions during that time. The genetic identity of one blood strain and all of the colonizing strains examined and the geographic and temporal associations are strong arguments in favor of considering this occurrence a definite cluster. No C. tropicalis isolates were identified in the unit during the last months of surveillance or subsequently. We suggest that nosocomial acquisition through indirect neonatal contact occurs, possibly via transient hand colonization of personnel. The use of molecular diagnostic tests, such as RFLP and PCR-based methods, is expected to provide more accurate means by which to understand the epidemiology of Candida infection, including the mode of transmission, and assist in the management of invasive fungal infections in the neonatal ICU. In conclusion, a substantial risk of nosocomial colonization by non-C. albicans Candida spp. in the neonatal ICU leads to a preponderance of C. parapsilosis and C. tropicalis as causes of neonatal fungemia. Nosocomial transmission of a predominant genotype of C. tropicalis may occur among neonates, probably via cross-colonization. Colonizing C. tropicalis strains appear to be capable of causing invasive disease in neonates; however, their tendency to progress from colonization to infection was found to be significantly lower than that of C. parapsilosis. ACKNOWLEDGMENTS The technical expertise and dedication of Avgi Maloukou is gratefully acknowledged. We also thank Zaharoula Topka, who performed the surveillance cultures in the neonatal ICU. REFERENCES 1. Baley, J. E., R. M. Kliegman, B. Boxerbaum, and A. A. Fanaroff. 1986. Fungal colonization in the very low birth weight infant. Pediatrics 78:225– 232. 2. Colombo, A. L., M. Nucci, R. Salomao, M. L. Branchini, R. Richtmann, A. Derossi, and S. B. Wey. 1999. High rate of non-albicans candidemia in Brazilian tertiary care hospitals. Diagn. Microbiol. Infect. Dis. 34:281–286. 3. da Silva, C. L., R. M. dos Santos, and A. L. Colombo. 2001. Cluster of Candida parapsilosis primary bloodstream infection in a neonatal intensive care unit. Braz. J. Infect. Dis. 5:32–36. 4. Fairchild, K. D., S. Tomkoria, E. C. Sharp, and F. V. Mena. 2002. Neonatal Candida glabrata sepsis: clinical and laboratory features compared with other Candida species. Pediatr. Infect. Dis. J. 21:39–43. 5. Faix, R. G., D. J. Finkel, R. D. Andersen, and M. K. Hostetter. 1995. Genotypic analysis of a cluster of systemic Candida albicans infections in a neonatal intensive care unit. Pediatr. Infect. Dis. J. 14:1063–1068.

J. CLIN. MICROBIOL. 6. Finkelstein, R., G. Reinhertz, N. Hashman, and D. Merzbach. 1993. Outbreak of Candida tropicalis fungemia in a neonatal intensive care unit. Infect. Control Hosp. Epidemiol. 14:587–590. 7. Flynn, P. M., N. M. Marina, G. K. Rivera, and W. T. Hughes. 1993. Candida tropicalis infections in children with leukemia. Leuk. Lymphoma 10:369–376. 8. Girmenia, C., P. Martino, F. De Bernardis, G. Gentile, M. Boccanera, M. Monaco, and A. Antonucci. 1996. Rising incidence of Candida parapsilosis fungemia in patients with hematologic malignancies: clinical aspects, predisposing factors, and differential pathogenicity of the causative strains. Clin. Infect. Dis. 23:506–514. 9. Horn, R., B. Wong, T. E. Kiehn, and D. Armstrong. 1985. Fungemia in a cancer hospital: changing frequency, earlier onset, and results of therapy. Rev. Infect. Dis. 7:646–655. 10. Huang, Y., T. Lin, H. Peng, J. Wu, H. Chang, and H. Leu. 1998. Outbreak of Candida albicans fungaemia in a neonatal intensive care unit. Scand. J. Infect. Dis. 30:137–142. 11. Huang, Y. C., R. I. Lin, Y. H. Chou, C. Y. Kuo, P. H. Yang, and W. S. Hsieh. 2000. Candidaemia in special care nurseries: comparison of albicans and parapsilosis infection. J. Infect. 40:171. 12. Huang, Y. C., T. Y. Lin, H. S. Leu, H. L. Peng, J. H. Wu, and H. Y. Chang. 1999. Outbreak of Candida parapsilosis fungemia in neonatal intensive care units: clinical implications and genotyping analysis. Infection 27:97–102. 13. Kao, A. S., M. E. Brandt, W. R. Pruitt, L. A. Conn, B. A. Perkins, D. S. Stephens, W. S. Baughman, A. L. Reingold, G. A. Rothrock, M. A. Pfaller, R. W. Pinner, and R. A. Hajjeh. 1999. The epidemiology of candidemia in two United States cities: results of a population-based active surveillance. Clin. Infect. Dis. 29:1164–1170. 14. Kossoff, E. H., E. S. Buescher, and M. G. Karlowicz. 1998. Candidemia in a neonatal intensive care unit: trends during fifteen years and clinical features of 111 cases. Pediatr. Infect. Dis. J. 17:504–508. 15. Krcmery, V., and A. J. Barnes. 2002. Non-albicans Candida spp. causing fungaemia: pathogenicity and antifungal resistance. J. Hosp. Infect. 50:243–260. 16. Leung, A. Y., C. S. Chim, P. L. Ho, V. C. Cheng, K. Y. Yuen, A. K. Lie, W. Y. Au, R. Liang, and Y. L. Kwong. 2002. Candida tropicalis fungaemia in adult patients with haematological malignancies: clinical features and risk factors. J. Hosp. Infect. 50:316–319. 17. Levy, I., L. G. Rubin, S. Vasistha, V. Tucci, and S. K. Sood. 1998. Emergence of Candida parapsilosis as the predominant species causing candidaemia in children. Clin. Infect. Dis. 26:1086. 18. Loudon, K. W., and J. P. Burnie. 1995. Comparison of three typing methods for clinical and environmental isolates of Aspergillus fumigatus. J. Clin. Microbiol. 33:3362–3363. 19. Lupetti, A., A. Tavanti, P. Davini, E. Ghelardi, V. Corsini, I. Merusi, A. Boldrini, M. Campa, and S. Senesi. 2002. Horizontal transmission of Candida parapsilosis candidemia in a neonatal intensive care unit. J. Clin. Microbiol. 40:2363–2369. 20. Makhoul, I. R., I. Kassis, T. Smolkin, A. Tamir, and P. Sujov. 2001. Review of 49 neonates with acquired fungal sepsis: further characterization. Paediatrics 107:61. 21. Narang, A., P. B. Agrawal, A. Chakrabarti, and P. Kumar. 1998. Epidemiology of systemic candidiasis in a tertiary care neonatal unit. J. Trop. Pediatr. 44:104–108. 22. National Committee for Clinical Laboratory Standards. 1997. Reference method for broth dilution antifungal susceptibility testing of yeasts. Approved standard M27-A. National Committee for Clinical Laboratory Standards, Villanova, Pa. 23. Powderly, W. G., G. S. Kobayashi, G. S. Herzig, and G. Medoff. 1988. Amphotericin B-resistant yeast infection in severely immunocompromised patients. Am. J. Med. 84:826–832. 24. Rangel-Frausto, M. S., T. Wiblin, H. M. Blumberg, L. Saiman, J. Patterson, M. Rinaldi, M. Pfaller, J. E. Edwards, W. Jarvis, J. Dawson, and R. P. Wenzel. 1999. National epidemiology of mycoses survey (NEMIS): variations in rates of bloodstream infections due to Candida species in seven surgical intensive care units and six neonatal intensive care units. Clin. Infect. Dis. 29:253–258. 25. Reagan, D. R., M. A. Pfaller, R. J. Hollis, and R. P. Wenzel. 1995. Evidence of nosocomial spread of Candida albicans causing bloodstream infection in a neonatal intensive care unit. Diagn. Microbiol. Infect. Dis. 21:191–194. 26. Reef, S. E., B. A. Lasker, D. S. Butcher, M. M. McNeil, R. Pruitt, H. Keyserling, and W. R. Jarvis. 1998. Nonperinatal nosocomial transmission of Candida albicans in a neonatal intensive care unit: prospective study. J. Clin. Microbiol. 36:1255–1259. 27. Roilides, E., I. Kadiltsoglou, D. Zahides, and E. Bibashi. 1997. Invasive candidiasis in pediatric patients: a prospective study. Clin. Microbiol. Infect. 3:192–197. 28. Saiman, L., E. Ludington, J. D. Dawson, J. E. Patterson, S. Rangel-Frausto, R. T. Wiblin, H. M. Blumberg, M. Pfaller, M. Rinaldi, J. E. Edwards, R. P. Wenzel, and W. Jarvis. 2001. Risk factors for Candida species colonization of neonatal intensive care unit patients. Pediatr. Infect. Dis. J. 20:1119–1124. 29. Saxen, H., M. Virtanen, P. Carlson, K. Hoppu, M. Pohjavuori, M. Vaara, J. Vuopio-Varkila, and H. Peltola. 1995. Neonatal Candida parapsilosis outbreak with a high case fatality rate. Pediatr. Infect. Dis. J. 14:776–781.

VOL. 41, 2003 30. Sullivan, D., D. Bennett, M. Henman, P. Harwood, S. Flint, F. Mulcahy, D. Shanley, and D. Coleman. 1993. Oligonucleotide fingerprinting of isolates of Candida species other than C. albicans and of atypical Candida species from human immunodeficiency virus-positive and AIDS patients. J. Clin. Microbiol. 31:2124–2133. 31. Viscoli, C., C. Girmenia, A. Marinus, L. Collette, P. Martino, B. Vandercam, C. Doyen, B. Lebeau, D. Spence, V. Krcmery, B. De Pauw, and F. Meunier. 1999. Candidemia in cancer patients: a prospective, multicenter surveillance study by the Invasive Fungal Infection Group (IFIG) of the European Organization for Research and Treatment of Cancer (EORTC). Clin. Infect. Dis. 28:1071–1079.

C. TROPICALIS IN A NEONATAL ICU

741

32. Walsh, T. J., and W. G. Merz. 1986. Pathologic features in the human alimentary tract associated with invasiveness of Candida tropicalis. Am. J. Clin. Pathol. 85:498–502. 33. Welbel, S. F., M. M. McNeil, R. J. Kuykendall, T. J. Lott, A. Pramanik, R. Silberman, A. D. Oberle, L. A. Bland, S. Aguero, M. Arduino, S. Crow, and W. R. Jarvis. 1996. Candida parapsilosis bloodstream infections in neonatal intensive care unit patients: epidemiologic and laboratory confirmation of a common source outbreak. Pediatr. Infect. Dis. J. 15:998–1002. 34. Wingard, J. R., W. G. Merz, and R. Saral. 1979. Candida tropicalis: a major pathogen in immunocompromised patients. Ann. Intern. Med. 91: 539–543.