Prophylactic Fluconazole Is Effective in Preventing Fungal Colonization and Fungal Systemic Infections in Preterm Neonates: A Single-Center, 6-Year, Retrospective Cohort Study

Study Design

A retrospective, nonrandomized intervention study with historical control subjects was undertaken by reviewing the clinical and microbiologic records of all VLBW neonates who were admitted to our NICU in the period 1998–2003. Those who were born in the period 1998–2000 and did not receive fluconazole prophylaxis were compared with those who were born in the period 2001–2003, all of whom received fluconazole.

Population

The study was conducted at the Neonatology and Hospital NICU of the Sant'Anna Hospital (Turin, Italy). This is a level 3 unit located in the greater Turin area (1500000 inhabitants and 15000 births per year) with a mean delivery rate of 4000 per year and 400 admissions to its neonatal subintensive and intensive care unit. Neonates who were born from January 1, 1998, when all medical records of NICU patients were stored in computerized database, to December 31, 2003, and survived >3 days were included in the study.

All cases were extracted from our database and reviewed by examining their demographic, gestational, and perinatal data as well as antenatal risk factors (specifically those associated with maternal and fetal diabetes), septic episodes, type and duration of nutrition, clinical and microbiologic-culture results, laboratory data, treatments, and outcome. A check was also made to ensure that all neonates with a diagnosis of SFI displayed the microbiologic laboratory and clinical criteria required. Two groups were formed: (1) group A, neonates who were born in the period 1998–2000 when fluconazole prophylaxis was not used, and (2) group B, neonates who were born in the period 2001–2003, all of whom received prophylactic fluconazole. Neonate characteristics are illustrated in Table 1.

Other causes that might have affected the frequency or relative weight of factors that increased or decreased the risk for SFI were ruled out. In detail, infection control policies were not significantly different in our unit or in the hospital as a whole during the period and followed the criteria expressed in protocols produced and regularly checked by a dedicated Control of Nosocomial Infections Committee. Furthermore, the quarterly surveillance reports that were issued by this committee never disclosed either an increase of fungal isolates that might have been related to problems in infection control or any episodic increase in Candida parapsilosis isolates (often related to the spread of hospital-acquired fungal infections). Finally, there were no changes in the policies and protocols regarding the use of antenatal and neonatal antibiotics or steroids or neonatal H2-antagonists or in nutritional protocols and policy. Informed written parental consent was obtained before any investigation or treatment.

Definition of SFI

Microbiologically documented fungal infection (proven fungal infection) was defined as a positive culture (1) from blood (drawn from peripheral sites) or (2) from urine (collected by suprapubic sterile puncture or sterile bladder catheterization, with growth of >10000 fungal organisms/mL) or (3) from cerebrospinal fluid or (4) from intravascular catheter tip (only considered proof of microbiologically documented fungal infection in patients with previous peripheral colonization by the same species; otherwise, positivity was considered as high-risk site colonization).

Clinically documented fungal infection (presumed fungal infection) was defined as clinical worsening with septic features confirmed by laboratory data (increase in serum C-reactive protein levels [>15 mg/dL], with leukocytosis [>30000/mm³] and/or leukopenia [<6000/mm³] and/or neutrophilia and/or neutropenia [Mouzinho reference values³⁰] and/or thrombocytopenia [<50000/mm³]) despite aggressive broad-spectrum antibiotic therapy for at least 2 days, with no isolation of other microorganisms in any cultures in the last 2 days, in patients who were already heavily colonized by Candida species (≥3 colonization sites at the onset of sepsis or previous positive culture from intravascular catheter tip).

These diagnostic criteria implemented guidelines from international Consensus documents,^{16, 17} as well as the Italian Neonatology Society's Fungal Infections Task Force recommendations.¹⁸

Treatment of SFIs

All SFI episodes were treated with intravenous liposomal amphotericin B (LAmB) given at 2.5 (start) to 5.0 (steady state) mg/kg per day for 12- to 28-day courses, as decided by the physician in charge.^{15, 31} When diagnosing an episode, removal of central vascular catheters was the standard policy for the management of central intravascular lines.³² When SFI was presumed, fluconazole was suspended and LAmB was given empirically until the culture results were known.¹³ 5-Fluorocytosine was added in 5 cases with end-organ damage and 1 neonate received LAmB plus itraconazole. No agent-related adverse effects or reactions were recorded, and there were no discontinuances in fluconazole prophylaxis.

Fungal Isolation and Identification From Cultures

From 5 to 16 of the following cultures per neonate were obtained:

• Cultures of surveillance (1 at day of life [dol] 1, 7, 14, 21, and 28 and also at dol 35 and 42 in ELBW infants): ear canal swab at birth, and then weekly stool, gastric aspirate, and nasopharynx secretions (endotracheal if intubated) cultures

• Cultures from surgical and mechanical devices when removed: endotracheal tubes and intravascular catheter, as well as from drains and similar devices

• Cultures from any site indicated by the physician and justified clinically (eg, skin, respiratory secretions).

Any surveillance cultures collected after dol 30 (dol 45 in ELBW infants) were not taken into consideration to exclude cases with unusually complicated courses. Baseline fungal colonization was defined as (1) ear canal swab at birth positive for fungi or (2) isolation of fungi from any site during dol 1 to 2. Low- and high-grade colonizations were defined as 1 to 2 and 3 to 4 sites colonized, respectively. Each site was considered once only, even when repeatedly positive. High-risk site colonization was defined as the isolation of fungi from urine (if <10000 organisms/mL), catheter tip (without contemporary peripheral colonization), and endotracheal tube; low-risk colonization was defined as the isolation from skin, stool, ear swab, gastric aspirate, and nasopharynx secretions.

Stool, gastric aspirates, surgical and intravascular devices were collected in sterile containers; respiratory secretions were obtained with an infant mucus sterile extractor kit supplied with two 3.3-mm suction catheters (Vygon, Ecouen, France); skin, ear, and nasopharynx specimens were obtained on swabs (Labobasi, Novazzano, Switzerland); blood draws for culture were submitted in dedicated specimens (BacT/Alert PF; BioMerieux Inc, Durham, NC). Urine samples were obtained by sterile urethral catheterization or suprapubic aspiration of the bladder; samples that were collected from indwelling catheters or from urine bags were not considered.

For the identification of fungi, all specimens were inoculated onto chromogen culture plates (Albicans ID; BioMerieux Inc), which allow for rapid Candida albicans identification through the blue staining of the colonies after 48 hours of incubation at 37°C. Differently stained colonies were speciated through a miniaturized system of biochemical tests (Vitec Yeast; BioMerieux Inc). The surveillance and culture collection and analysis procedures and methods were not changed at any time during the period studied.

Prophylaxis With Fluconazole in Group B Neonates

The group B regimen was (and still is) 6 mg/kg fluconazole (Diflucan; Pfizer Italia s.r.l., Latina/Roma, Italy) every 72 hours in the first week of life, then every 48 hours from the second week until dol 30 for NE-VLBW and dol 45 for ELBW neonates or until earlier discharge or until the need for systemic antifungal therapy (mostly with LAmB) as a result of the onset of proven or presumed SFI. Fluconazole was administered starting from dol 1 as a single dose intravenously or orally, depending on the availability of a venous line and/or on the tolerance of oral feeding. This schedule was based on experiences and published pharmacokinetic data from preterm neonates.^{24, 33–36}

Routine surveillance (serum alanine aminotransferase and aspartate aminotransferase concentrations checked twice) was performed to record the presence of drug-related adverse effects or toxicity. Neonates were also screened for interactions between fluconazole and other concomitantly prescribed drugs and for any clinical manifestation that might be related to the use of fluconazole.

Patterns of Sensitivity of Candida Species to Fluconazole During the Observation Period

Fungal isolates were tested for susceptibility to fluconazole at the hospital's microbiology service at the time of isolation. The standardized microbroth dilution assay was used according to the National Committee for Clinical Laboratory Standards standard recommendations.³⁷ The interpretative breakpoints of fluconazole resistance were defined as ≥64 μg/mL, as recommended by the National Committee for Clinical Laboratory Standards.³⁷

Statistical Analysis

The following variables were analyzed in both groups and for ELBW and NE-VLBW separately:

• Incidence of colonization during the first month of stay in NICU;

• Incidence of baseline colonization;

• Incidence of low- and high-grade colonization;

• Incidence of low- and high-risk site colonization;

• Incidence of SFI (total, proven, and presumed);

• Overall mortality rate expressed as proportions (mortality for all causes before hospital discharge);

• Mortality rate (expressed as proportions) attributable to fungal infections: defined as death within 3 days from the last positive culture from a high-risk site in the absence of any other cause of death or as isolation of Candida from autopsy specimens;

• Rate of progression from colonization to systemic infection (expressed as proportions);

• Incidence of isolation of natively fluconazole-resistant fungal species;

• Patterns of sensitivity to fluconazole of fungal isolates; and

• Incidence of secondary outcomes in group B, particularly those potentially related to the use of fluconazole.

Yates' corrected χ² to compare proportions and relative risk (RR; risk ratio) estimates, differences in risk to compare incidence rates among groups, and t test for continuous variables were calculated with SPSS 8.0 for Windows statistical software (SPSS Inc, Chicago, IL). Power calculations were performed according to S Plus 2000 (MathSoft, Cambridge, MA).

Population

Neonate characteristics are illustrated in Table 1. The records showed that 490 VLBW neonates (all inborn) had been admitted to our NICU and had survived >3 days. Twenty-five (12 in group A, 13 in group B) were excluded because of 1 of the following: incomplete or unavailable data, incorrect prophylaxis, ineligibility (1 group B neonate with abnormal serum liver enzyme levels on enrollment), or absence of parental consent. None of the 3 neonates who were born to HIV-positive mothers was HIV-positive at 1 year of age. One had received prophylactic fluconazole during the period 1998–2000 and therefore was excluded. The other 2 were treated in the period 2001–2003 and were free from colonization and infection by fungi. The remaining 465 neonates were considered eligible for the study and divided into 2 groups: (1) group A, neonates who were born in the period 1998–2000, when fluconazole prophylaxis was not used (n = 240: 57 ELBW, 183 NE-VLBW), and (2) group B, neonates who were born in the period 2001–2003, all of whom received prophylactic fluconazole (n = 225: 72 ELBW, 153 NE-VLBW).

There were no significant differences at the baseline between groups A and B in terms of presence of major (including antenatal) risk factors for fungal colonization and systemic infection, as defined by most reports.^1–10 Mean birth weight was significantly lower in group B (1108 vs 1212 g; P = .001), in keeping with the tendency to admit more complicated pregnancies and deliveries to our department. For confirming the importance of the results, the a posteriori power for subgroup analysis was .97 in the ELBW group and .11 for the NE/VLBW group.

Mortality

See Tables 1 and 2. Overall mortality was similar in both groups (11.2% in group A vs 10.6% in group B; P = .44). In colonized infants (n = 159), it was significantly lower in group B (3.7% vs 18.1%; 95% confidence interval [CI]: 0.039–0.778; P = .007). The decrease was significant in ELBW (10.5% vs 39.0%; 95% CI: 0.037–0.905; P = .01) but not in NE-VLBW colonized neonates (3.0% vs 4.7%; P = .26). Mortality attributable to Candida was 0% (0 of 225) in group B (instead of the 1.9% expected) vs 1.7% (4 of 240) in group A (RR: 0.146; P = .07)

Incidence of SFI

The incidence of SFI was significantly lower in group B than in group A (P < .0001), both in ELBW and in NE-VLBW neonates (P < .0001 and P = .02, respectively). Significance was also maintained (P = .001) when the proven episodes that were diagnosed from catheter tips (5 of 27 in group A, 1 of 5 in group B) were excluded. The difference in presumed SFI was not significant between groups (P = .06).

We also analyzed the ELBW group further, examining the infants who weighed <750 g (n = 49) and the infants who weighed 750 to 999 g (n = 80), for the presence of SFI. Data show that reduction in SFI obtained with fluconazole was significant in both subgroups (P < .0001 in infants <750 g, P = .02 in infants 750–999 g, respectively).

There were 4 cases of fatal SFI (10% of SFI episodes), all in group A (3 in ELBW neonates): 2 were proven episodes, 1 (in an ELBW patient) was presumed, and the fourth was presumed and became proven after autopsy. C albicans was detected in all 4 fatal cases. There were no fatal cases in group B (P = .07), instead of the 1.9 expected. No clusters of SFI occurred. In 11 cases of SFI, neutropenia prompted the use of short courses of filgrastim (granulocyte colony-stimulating factor; always before detection of SFI).

The 50 episodes of SFI were caused by C albicans (42 cases), C parapsilosis,⁶Candida glabrata,⁴Candida krusei,²Aspergillus fumigatus,¹Candida tropicalis,¹ and Candida guilliermondii.¹ Seven neonates were infected by 2 species, but neither was a natively fluconazole-resistant fungal species.

Incidence of Colonization

Baseline colonization rate was similar: 3.7% in group A vs 4.9% in group B (P = .15). Prophylactic fluconazole decreased all colonization rates (overall, low-grade, high-grade, low-risk, and high-risk colonization rate) in group B neonates. Clustering the analysis for weight range, prophylactic fluconazole caused an equally significant decrease of colonization in ELBW (from 71.9% to 26.4%; P < .0001) and NE-VLBW neonates (from 35.0% to 22.2%; P = .01; Table 3).

Relationship Colonization/SFI

Statistical analysis showed an association between the occurrence of SFI and the presence of colonization. The infection/colonization ratio was lower in group B (0.17) than in group A (0.38), showing that fluconazole reduced or prevented the progression to infection in more than half of the colonized neonates in group B (RR: 0.369; 95% CI: 0.159–0.815). This trend was unchanged when proven and presumed SFI were analyzed separately. Fluconazole was most effective in reducing or preventing progression from colonization to infection in neonates with low-grade colonization and in those with low-risk colonization (P = .002 in both cases), regardless of their birth weight, and in ELBW infants as a whole (P = .009), whereas it was much less effective in neonates who were already heavily colonized (P = .08) or colonized in high-risk sites (P = .15; Table 4).

Incidence of Fluconazole-Resistant Fungal Species Isolates

There were no significant between-group differences in the incidence of potential, demonstrated, or natively fluconazole-resistant species such as C krusei or C glabrata (P = .51). The number of isolates from these species was very low in group B (4 C krusei and 5 C glabrata). More important, the overall crude number of isolates from natively fluconazole-resistant species was 14 in group A vs 15 in group B, indicating the absence of a shift toward resistant species (the place left vacant by the nonresistant species was still vacant after 3 years). In addition, the number of episodes that were caused by C krusei and C glabrata was 4 in group A and 3 in group B (RR: 0.895; 95% CI: 0.456–1.320), and none was fatal. Mortality rate for nonfungal causes in this subgroup was lower in group B (0 of 9 vs 2 of 9; P = .20).

Patterns of Sensitivity of Candida Species to Fluconazole During the Observation Period

Sensitivity to fluconazole of fungal isolates (as expressed by minimal inhibitory concentration for 90% of colonies [MIC90] values) did not change significantly, and all isolates remained sensitive to fluconazole during the first 3 years and at the fourth and sixth years (MIC [μg/ml] of C albicans: 0.125–2.0; of C parapsilosis: 1.0–4.0; and of C glabrata: 2–16; in this last case, the value 16 is indicative of dose-dependent susceptibility rather than true sensitivity). The highest recorded values of MIC90 of fluconazole (expressed as μg/ml) for the fungal subspecies for whom data allowed analysis, during the 6-year survey, were 2.0 in group A and 1.0 and 2.0 in group B (at the first and third years of prophylaxis, respectively) for C albicans and 4.0 in group A and 4.0 and 2.0 in group B (at the first and third years of prophylaxis, respectively) for C parapsilosis.

Six-Year Time Trend

The 6-year time trends of incidence of colonization, SFI, and progression rate (Table 5) show that an abrupt drop in all 3 parameters appeared as early as the first year of prophylactic fluconazole.

Secondary Outcomes

There were no significant differences in secondary outcomes between the 2 groups. The incidence of gastroesophageal reflux, bronchopulmonary dysplasia, intraventricular hemorrhage, major surgery (eg, ligation of patent ductus arteriosus) was not significantly different. A point of great importance is that the frequency of bacterial infections (particularly those by Staphylococcus epidermidis) and the incidence of necrotizing enterocolitis (both surgical and nonsurgical) did not change with the use of fluconazole. Only the frequency of grades 3 and 4 retinopathy of prematurity was significantly lower in group B.

There were no adverse effects or toxicity related to fluconazole. There was never an unacceptably high increase in serum hepatic enzymes or liver function impairment or clinical signs of hepatotoxicity. Mean serum values of aspartate aminotransferase and alanine aminotransferase were not different between the 2 groups (see Table 1). Incidence of hyperbilirubinemia was the same in both groups. Fluconazole administration was never discontinued on account of adverse effects or intolerance.