Patterns of Fungal Colonization in Preterm Infants Weighing Less Than 1000 Grams at Birth

Abstract

Background: 

Colonization with Candida spp. is an important risk factor for systemic infection in very low birth weight (VLBW; <1500 g) and extremely low birth weight (ELBW, <1000 g) infants. ELBW infants are at a higher risk than VLBW infants for fungal sepsis and its associated mortality, but few studies have examined fungal colonization exclusively in ELBW infants.

Methods: 

Fungal colonization data were analyzed retrospectively in 50 high risk ELBW infants. Weekly surveillance fungal cultures of the skin, gastrointestinal tract, respiratory tract and umbilicus had been performed from birth through the first 6 weeks of life. Colonization was analyzed for time of initial colonization, site, species and spread of Candida from one site to another.

Results: 

Candida was isolated from surveillance cultures in 31 of 50 (62%) infants. Colonization was inversely proportional to gestational age. Initial week of both the fungal colonization of the skin [1 (0–6) week, median (range)] and gastrointestinal tract [2 (0–6)] preceded colonization of the respiratory tract [3 (1–6)] (P = 0.0001). Among infants colonized by only 1 of the species, colonization at 2 or more sites occurred similarly with Candida albicans (77%) and Candida parapsilosis (85%), whereas colonization at 3 or more sites occurred more frequently with C. albicans (69%) compared with C. parapsilosis (23%) (P = 0.047).

Conclusions: 

Fungal colonization occurs on the skin and gastrointestinal tract before the respiratory tract. In addition, C. albicans is more likely than C. parapsilosis to colonize multiple sites.

Fungal colonization is an important risk factor for systemic infection,¹ which leads to significant neurodevelopmental impairment and has a mortality rate of up to 32% in very low birth weight (VLBW; <1500 g) infants.^2–7 In a single center study of VLBW infants (N = 146), fungal colonization occurred in 26.7%, and one-third of those colonized developed disease.⁸ In a multicenter study of neonatal candidemia, molecular typing demonstrated that the same fungal isolate was present in the bloodstream and gastrointestinal tract, illustrating the relationship between gastrointestinal tract colonization and fungemia.¹ The same multicenter study found that rectal fungal colonization was a risk factor for sepsis by univariate analysis, but controlling for birth weight <1000 g and abdominal surgery, rectal colonization was not statistically significant.¹ That and other studies have reported that extremely low birth weight (ELBW; <1000 g) infants have additional factors that increase their risk for fungal sepsis.¹

Preterm infants weighing <1000 g at birth are at a higher risk for candidemia and fungal-associated mortality than those infants with 1000–1500 g birth weight.^1–3,9,10 Patterns of colonization also differ between infants <1000 g and those 1000–1500 g, but few studies have addressed this issue. In the multicenter study of rectal fungal colonization discussed above, infants >1000 g were more likely to be colonized with Candida albicans than Candida parapsilosis, whereas in infants <1000 g there was no difference in the rate of C. albicans and C. parapsilosis colonization.¹¹

We previously reported similar findings with regard to species colonization in ELBW infants.⁹ In addition, we have reported a correlation between the risk of candidemia and the number of sites colonized in ELBW infants.⁹ In this study, we further analyze patterns of fungal colonization in the ELBW infant, including timing, sites, species, birth weight and gestational age of Candida colonization.

MATERIALS AND METHODS

Study Design

We collected fungal surveillance cultures as part of a prospective randomized double blind clinical trial that evaluated the efficacy of fluconazole prophylaxis in preterm infants.⁹ To examine fungal colonization patterns in ELBW infants, we analyzed the fungal colonization patterns of the 50 patients in the placebo group. We did not include any of the fluconazole-treated patients, given that antifungal treatment significantly decreased fungal colonization and this would affect the patterns of colonization analyzed in this study. All infants admitted to our neonatal intensive care unit were eligible, whether born at our institution or transferred from another facility, if they had a birth weight of <1000 g, were younger than 5 days old and had a central vascular catheter, an endotracheal tube or both. Before enrollment, written informed consent was obtained from each infant's parent or guardian. The protocol was approved by the University of Virginia Human Investigation Committee.

Patient data were prospectively collected during the subjects’ hospitalization and through review of their medical records.

Fungal Isolation and Identification

Fungal surveillance cultures of the skin (intertriginous area), gastrointestinal tract (rectal swab or stool) and respiratory tract (tracheal secretions if intubated and nasopharyngeal sample if not intubated) were obtained by 1 clinical nurse or the principal investigator associated with the study at baseline and then weekly during the 6-week study period on the day of the week of the patient's enrollment. Umbilical cultures were also performed at baseline and week 1.

Specimens were obtained on swabs (Culturette; Becton Dickinson Microbiology Systems, Sparks, MD) except for stool and tracheal secretions which were submitted in sterile containers. Each specimen was inoculated onto Inhibitory Mold Agar (Remel, Lenexa, KS), and the plates were incubated and examined repeatedly for 4 weeks at 30°C incubation unless yeast growth occurred by 2 weeks.

Speciation of fungal isolates were done by standard methods, including an initial screen for germ tube formation and exoenzyme production (C. albicans screen; Carr-Scarborough Microbiologicals, Decatur, GA) and a commercial auxanographic method (API 20 C AUX; bioMérieux, Hazelwood, MO).^3,12

Statistical Analysis

All analyses used SAS version 9.1 (SAS Institute, Cary, NC). Several aspects of fungal colonization were examined, including time to initial colonization, comparison of colonization across sites, comparison of species and the impact of birth weight and gestational age on colonization. Comparisons of colonized infants versus noncolonized infants for a variety of characteristics were done by Fisher's exact tests (Table 1). Multiple logistic regression was used to assess the impact of birth weight and gestational age on odds of colonization. Exact methods for estimating the parameters of the logistic model were used. Next a Cox proportional hazards model was fit to compare time to initial colonization among sites (gastrointestinal tract, respiratory tract and skin). Because these sites are not independent within an individual, the generalized estimating equations approach for inference on the factors in the model was used.¹³ The Cox model included birth weight and gestational age as covariates. The proportional hazards assumption for each variable in the model was assessed using numerical methods based on the cumulative sums of martingale residuals.¹⁴

TABLE 1: Comparison of Colonized and Noncolonized Patients

We focused on comparing C. albicans and C. parapsilosis with regard to the number of sites colonized because few other species were isolated. For valid comparisons relying on the assumption of independence between the 2 groups of species, 3 infants who were colonized by both C. albicans and C. parapsilosis were not included in the analysis. Odds ratios were estimated, and corresponding χ² tests were computed. Fisher exact test was used to compare the proportions of interest.

RESULTS

Patient Characteristics

As reported in a randomized trial,⁹ the 50 placebo-treated patients studied had a birth weight (mean ± SD) of 744 ± 159 g and a gestational age of 25.6 ± 2.0 weeks. This included ∼90% of infants born weighing <1000 g during the study period. Infants were screened on average (±SD) for 5.2 ± 1.4 weeks. Risk factors for invasive fungal infection during the surveillance period included parenteral nutrition (N = 100%), umbilical catheters (100%), endotracheal intubation (100%), central venous catheters (96%), postnatal steroids (72%), third generation cephalosporins (68%) and histamine-2 receptor (H2) antagonists (22%). On average, infants were exposed to antibiotics 14 ± 8 (mean ± SD) days. Of these characteristics, cephalosporin exposure was greater in colonized (84%) than in noncolonized infants (42%) (P = 0.002) (Table 1).

Fungal Colonization

Candida was isolated from surveillance cultures of the skin, respiratory tract or gastrointestinal tract at least once in 31 of 50 infants (62%) at baseline or during the 6-week surveillance period. The species isolated included C. albicans (n = 16), C. parapsilosis (16), Candida glabrata (1), Candida lusitaniae (1), Trichosporon mucoides (1), and Candida guilliermondii (2). Three infants were colonized with 2 and 1 infant with 3 fungal species during the 6-week surveillance period.

Initial Colonization.

Initial colonization by week is presented in Figure 1. Initial colonizing species included C. albicans (n = 14), C. parapsilosis (13), C. glabrata (1), C. lusitaniae (1) and C. guilliermondii (1). Twenty-seven (87%) of the 31 infants were initially colonized within the first 2 weeks of life. After the first 2 weeks, 4 patients had their initial colonization with C. albicans (n = 3) and C. parapsilosis (n = 1). After initial colonization with the first fungal species, fungal colonization with a different species occurred during weeks 2 and 3 with C. albicans, C. parapsilosis, C. guilliermondii and T. mucoides.

FIGURE 1.: Initial week of fungal colonization.

C. albicans and C. parapsilosis colonization initial colonization is shown in Figure 2. Week of initial colonization was similar between species, with slightly more C. albicans colonization present at baseline (5 of 16) compared with C. parapsilosis (2 of 16) (P = 0.39).

FIGURE 2.: Initial colonization by Candida species.

Length of Colonization.

During the surveillance period, 18 infants remained colonized during consecutive weeks, and 13 infants had intermittent colonization during the weekly screening after their initial week of colonization. Infants were colonized with C. albicans for a median (range) of 3 (1–7) weeks and C. parapsilosis for 4 (1–6) weeks.

Colonization by Site.

The proportional hazards assumption was not found to be violated for any of the covariates in the Cox model. Model-estimated survivor curves in Figure 3 show the differences among sites in probability of being colonized. Because gestational age and birth weight were included in the Cox model, survivor function estimates from the model used the sample mean value for both variables (744 g and 26 weeks, respectively). The skin and gastrointestinal tract had significantly higher hazards of colonization than the respiratory tract (P = 0.003 and P < 0.001, respectively). This is reflected by the fact that the week of initial fungal colonization skin and gastrointestinal tract preceded initial respiratory tract colonization. There was no significant difference between the hazards of skin and gastrointestinal tract colonization (P = 0.081). A significant inverse relationship between hazard of initial colonization and gestational age was found (P < 0.001). Birth weight had no effect in the model.

FIGURE 3.: Fungal colonization by site over time. Cox proportional hazards model-estimated survivor curves for colonization by site over time. Displayed model-estimated survivor curves for an infant with a birth weight of 744 g and gestational age of 26 weeks (sample means from the dataset). Skin and gastrointestinal tract colonization occurred prior to colonization of the respiratory tract. Overall significant difference among the 3 sites (P = 0.0001).

C. albicans was isolated on the skin in 56% (9 of 16) of colonized infants, and colonization was present in 56% (5 of 9) of the infants at the initial week of colonization. C. albicans gastrointestinal colonization was detected in 94% (15 of 16) of colonized infants, and colonization was present in 80% (12 of 15) of these infants at the initial week of colonization.

For C. parapsilosis, skin colonization was detected in 88% (14 of 16) of colonized infants, and colonization was present in 57% (8 of 14) of the infants at the initial week of colonization. C. albicans gastrointestinal colonization was detected in 75% (12 of 16) of colonized infants and colonization was present in 83% (10 of 12) of these infants at the initial week of colonization.

C. albicans was isolated from the respiratory tract of 63% (10 of 16) colonized infants, but colonization was present in only 20% (2 of 10) of these infants at the week of initial fungal colonization. In 60% (6 of 10) of these infants, gastrointestinal colonization with C. albicans preceded respiratory tract colonization by 1–2 weeks, and in 30% (3 of 10) skin colonization preceded respiratory colonization by 1 week. C. parapsilosis was isolated from the respiratory tract in 44% (7 of 16) colonized infants.

Multiple Site Colonization.

Colonization at 2 or more sites was similar between C. albicans and C. parapsilosis (P = 1.00), while colonization at 3 or more sites more commonly occurred with C. albicans (P = 0.047) (Table 2).

TABLE 2: Multiple Site Colonization by Species

Colonization by Gestational Age and Birth Weight.

Odds of Candida colonization was modeled with the use of multiple logistic regression with birth weight and gestational age as covariates. The odds of Candida colonization were inversely proportional to gestational age (P = 0.009), but not significantly associated with birth weight (P = 0.66).

DISCUSSION

In this study of high risk ELBW infants, we found that C. albicans was more likely than C. parapsilosis to colonize multiple sites and that fungal colonization of the skin and gastrointestinal tract occurred before that of the respiratory tract. Similar to studies of fungal colonization in VLBW infants, the majority of colonization in high risk ELBW infants occurred within the first 2 weeks of life.¹⁵

Although both C. albicans and C. parapsilosis colonized multiple sites, C. albicans was more likely to colonize 3 or more sites. This is important because the number of sites colonized correlates with risk of infection in ELBW infants. It is possible that the burden of fungi is greater with C. albicans; and this may contribute to poorer outcomes given that mortality associated with C. albicans sepsis is 3 times greater than that with C. parapsilosis sepsis in VLBW infants.^2,9,16 This may also imply increased adherence and proliferation by C. albicans compared with C. parapsilosis and other Candida species. In our previous study,⁹ multivariate analysis did not demonstrate that colonization with specific Candida species or site was associated with an increased risk for invasive fungal infection or mortality.

Candida colonization of the skin and gastrointestinal tract preceded respiratory tract colonization. Curiously ELBW infants are usually intubated at birth, but the respiratory tract was not colonized until an average of 3 weeks of life. Candida adheres well to plastics, but this delayed colonization implies that fungus may migrate from the skin or gastrointestinal tract to the respiratory tract via the epithelium and mucosal membranes or spread by health care workers or family members.¹⁷ Preventing respiratory fungal colonization may be a point where prevention of invasive disease in intubated preterm infants could occur; Rowen et al¹⁸ found that endotracheal colonization significantly increased the risk for systemic infection compared with colonization at other sites in VLBW infants.

Studies have reported that Candida colonization of the respiratory tract to be more common in intubated ELBW infants than in intubated VLBW infants.^9,18Candida respiratory tract colonization in intubated ELBW infants was almost 4 times (42% versus 11%) the incidence and more than twice the rate of C. albicans respiratory tract colonization (20% versus 8.6%) than intubated VLBW infants.^9,18 This may be due to the more immature immune system and prolonged duration of intubation in ELBW compared with VLBW infants.¹⁹

Some studies have used the presence of fungal colonization to identify high risk ELBW infants for prophylactic measures. Unfortunately, this strategy has been unsuccessful using nystatin, possibly due to the fungal burden present when colonization is detected.^20–22 This may also be the result of the site(s) or timing of fungal screening cultures. Because the gastrointestinal tract is the most common site of colonization, it is often the site chosen. Gastrointestinal tract colonization was present the initial week of colonization in 71% (22 of 31) of infants and 87% of patients during the 6 weeks of our study; therefore not all colonized infants can be identified this way. In this and other studies, the initial colonization of ELBW infants occurred most often in the first 2 weeks of life.^15,23 This is the same period when approximately one-half of late onset infections occur, therefore waiting until colonization is identified is often too late to institute preventative strategies.^2,9,15,23 Strategies aimed at prevention must occur before identifying initial colonization.

The major limitation of the study is that isolates were not analyzed by molecular typing. We made the assumption that the same strain of Candida was the colonizing strain and was present week to week and at different colonization sites. This study has relatively small number of subjects and was done at a single center; therefore variability in fungal exposure, clinical and infection control practices affecting the rates of fungal colonization and invasive disease may be different from those in other neonatal intensive care units. One example is that the effect of third generation cephalosporin use was more common in the colonized patients and may have affected colonization rates. Finally these results were obtained in high risk ELBW infants, and fungal colonization may be different in patients with fewer risk factors for infection.

The ELBW infant differs from the VLBW infant in incidence and patterns of fungal colonization. Because colonization often precedes infection, its prevention is important in decreasing serious fungal disease in ELBW infants. Infection control strategies that would prevent colonization, spread and proliferation of fungal colonization are important in the care of these infants, because these critically ill high risk patients in one study had close to 40 direct touches per shift with only 20% with appropriate washing or gloving before contact.^24,25 Clinical practices also play a role given that exposure of cephalosporins and H2 antagonists are risk factors for fungal colonization and fluconazole prophylaxis decreases fungal colonization.^{1,9,11,23,26,27} Studies of fluconazole prophylaxis have been effective in preventing invasive fungal disease^9,26–28 and other strategies affecting fungal colonization may lead to decreased disease.