Abstract
Background. Colonization by Candida species is the most important predictor of the development of invasive fungal disease in preterm neonates, and the enteric reservoir is a major site of colonization. We evaluated the effectiveness of an orally supplemented probiotic (Lactobacillus casei subspecies rhamnosus; Dicoflor [Dicofarm spa]; 6 × 109 cfu/day) in the prevention of gastrointestinal colonization by Candida species in preterm, very low birth weight (i.e., <1500-g) neonates during their stay in a neonatal intensive care unit.
Methods. Over a 12-month period, a prospective, randomized, blind, clinical trial that involved 80 preterm neonates with a very low birth weight was conducted in a large tertiary neonatal intensive care unit. During the first 3 days of life, the neonates were randomly assigned to receive either an oral probiotic added to human (maternal or pooled donors') milk (group A) or human milk alone (group B) for 6 weeks or until discharge from the NICU, if the neonate was discharged before 6 weeks. On a weekly basis, specimens obtained from various sites (i.e., oropharyngeal, stool, gastric aspirate, and rectal specimens) were collected from all patients for surveillance culture, to assess the occurrence and intensity of fungal colonization in the gastrointestinal tract.
Results. The incidence of fungal enteric colonization (with colonization defined as at least 1 positive culture result for specimens obtained from at least 1 site) was significantly lower in group A than in group B (23.1% vs. 48.8%; relative risk, 0.315 [95% confidence interval, 0.120–0.826]; P = .01). The numbers of fungal isolates obtained from each neonate (P = .005) and from each colonized patient (P = .005) were also lower in group A than in group B. L. casei subspecies rhamnosus was more effective in the subgroup of neonates with a birth weight of 1001–1500 g. There were no changes in the relative proportions of the different Candida strains. No adverse effects potentially associated with the probiotic were recorded.
Conclusions. Orally administered L. casei subspecies rhamnosus significantly reduces the incidence and the intensity of enteric colonization by Candida species among very low birth weight neonates.
Candida species are the third most frequent causal agent of late-onset sepsis in preterm neonates [1–3], with estimated incidences of 1.6%–9% among very low birth weight neonates and 15% among extremely low birth weight neonates in neonatal intensive care units (NICUs) [2, 4]. The frequency of invasive fungal infections (IFIs) is reported to have steadily increased during the past 2 decades [2, 3]; systemic disease caused by Candida species is currently a major threat to the survival of premature neonates, because it is associated with substantial morbidity and neurodevelopmental impairment [5], as well as a high attributable mortality rate (range, 30%–75%) [1, 5]. Preterm neonates in NICUs are indeed highly prone to develop IFI, because of a mostly unavoidable presence of a wide pattern of risk factors that have been extensively investigated [1–5]. Of these risk factors, colonization by Candida species is the most important predictor of invasive disease. Fungal colonization is the condition most frequently reported to be associated with an increased risk of development of IFI [4–10], and if careful investigation is done, it can always be found preliminarily to the development of any IFI [11–13]. It is estimated that 60% of very low birth weight neonates become colonized by fungi during the first month of life in the NICU, and many of them become infected after having been colonized. It has been estimated that, of 100 very low birth weight neonates in a NICU, 33 develop fungal colonization and 7 of them have colonization progressing to IFI [6]. Of all colonization sites [6–9, 12, 13], the gastrointestinal tract seems to be the site that is most frequently implicated in subsequent systemic dissemination [14], and some investigators have considered a positive rectal culture result to be an expression of fungal colonization [15].
Achieving a reduction in rates of fungal colonization has been proven to be effective in preventing IFI. Systemic antifungal drugs have been tested and have shown promising results [15–18], but prophylaxis with antifungal agents still raises concerns about tolerability and potential selection of resistant strains, and it is not yet implemented as a standard of care [19–21].
Probiotics modify the enteric microflora by colonizing the gastrointestinal tract, and they reduce the overgrowth of pathogens in the gut of the preterm neonate [22–24], as well as the incidences of necrotizing enterocolitis [25–30] and bacterial sepsis [26]. For fungi as well, studies of mice models have shown the ability of Lactobacillus casei subspecies rhamnosus (LGG) in reducing both enteric colonization and fungal systemic infections [31, 32].
Our hypothesis was that increased colonization with desirable microflora, such as the probiotic LGG, would protect the neonate host from the expansion of fungal colonies in the gastrointestinal tract [25, 33]. We tested this hypothesis by performing a clinical trial involving preterm, very low birth weight neonates.
Population, Materials, And Methods
Study Design
During a 12-month period, we performed a prospective, randomized, double-blind clinical trial involving 80 preterm neonates with a birth weight of <1500 g who were admitted to our NICU. During the first 3 days of life, the neonates were randomly assigned to receive either LGG with human milk or human milk alone for 6 weeks or until discharge from the NICU, if discharge occurred before 6 weeks.
Setting
The study was conducted at the Neonatology and Hospital NICU of S. Anna Hospital (Turin, Italy), a level III neonatal center located in the greater Turin area (which has 1,500,000 inhabitants and where 15,000 births occur each year). The mean delivery rate at the hospital is 4000 births/year, and 400 admissions to the Neonatal Subintensive and Intensive Care Unit occur each year.
Methods and Population
All neonates who weighed <1500 g at birth, were admitted to our NICU, were >3 days of age, and had started oral feeding with human milk (either their own mother's milk or donor milk) before randomization were evaluated to determine their eligibility for inclusion in the study. The criteria for eligibility for randomization were as follows:
1. The neonate was born in S. Anna Hospital.
2. The neonate did not have baseline fungal colonization at enrollment (with colonization defined by an ear canal swab specimen obtained at birth and found to be positive for fungi or by isolation of fungi from a culture specimen obtained from any site during the first 3 days of life).
3. The neonate did not receive any form of antifungal prophylaxis other than LGG.
4. Oral feeding was stable and was tolerated by the neonate at the time of randomization.
5. Informed, written consent was provided by the parent(s) of the neonate.
Neonates eligible for inclusion in the study were randomly assigned to 2 different groups. Group A included neonates who received supplementation with an oral probiotic (LGG [Dicoflor 60; Dicofarm spa]; 6 × 109 cfu/day) administered from the third day of life until either the end of the sixth week of life or until discharge from the NICU, if it occurred before the sixth week of life. Group B included neonates who were not receiving LGG supplementation.
Randomization was performed by assigning the neonates to the 2 groups according to a computer-generated randomization table. Neonates in group A had oral probiotic administered with the human (donor or mother's) milk in a single dose (i.e., half of a packet of Dicoflor 60) during the day. It was prepared via a suspension of freeze-dried powder in sterile, distilled water, with a final concentration of 6 × 109 cfu/mL. LGG was added to the milk immediately after its preparation, according to randomization, by the staff at the Human Milk Bank facility at our institution, who prepare the feedings for all neonates admitted to the NICU on a daily basis. In any case, administration of LGG after reconstitution occurred in <2 h; the enteral feeding rate for all newborns was 8–12 meals/day, and LGG was always administered with the first or second meal. The staff at the Human Milk Bank, as well as the staff at the hospital's Service of Microbiology, were blinded to the clinical and microbiological histories of the neonates.
The dose of 6 × 109 cfu/day was chosen on the basis of published data from previous studies of very low birth weight neonates [26]. Neonates in group B received the prescribed volume of human milk without any addition of probiotic.
For all neonates enrolled in the study, a database was kept that collected all demographic, gestational, and perinatal data, as well as data about antenatal and perinatal risk factors for fungal colonization. Data on the type and duration of nutrition, clinical characteristics, microbiological culture results, laboratory results, treatments (specifically, treatment with antibiotics and systemic steroids), septic episodes, and outcome were recorded prospectively. Although the influence of LGG on the incidence of IFI was not the aim of the present study, neonates who had IFI diagnosed underwent additional evaluations in which their medical charts were reviewed for the presence of the microbiological, laboratory, and clinical criteria needed to fulfil this diagnosis.
On a weekly basis, oropharyngeal, stool, gastric aspirate, and rectal specimens for surveillance culture were obtained from enrolled patients on days of life 1, 7, 14, 21, 28, 35, and 42. Culture specimens that were obtained from any site indicated by the physician and were justified clinically, as well as culture specimens obtained from surgical and mechanical devices after removal of the devices, were obtained with nonstandardized frequency, but they were not considered for the study.
The intensity of colonization was evaluated by calculating both the mean number of fungal isolates obtained from each colonized patient and the relative proportions of neonates with low-grade and high-grade colonization in the 2 groups. Low-grade colonization was defined as colonization of 1 site, and high-grade colonization was defined as colonization of ⩾2 sites. Each site was considered only once, even if the cultures from the specimens obtained from that site were repeatedly positive.
During the study period, infection-control policies did not significantly differ in our unit, and they followed criteria expressed by protocols produced and regularly checked by the Hospital Committee for the Control of Nosocomial Infections. The quarterly surveillance reports released by the committee never revealed either an increase in the number of fungal isolates eventually associated with problems involving infection control or an episodic outbreak of infection due to Candida species eventually associated with the spread of hospital-acquired fungal infections. In addition, there was no change in the policies and protocols regarding the use of antenatal and neonatal antibiotics or steroids or the use of neonatal histamine-2 blockers.
Feeding and nutritional practice followed the same protocols during the study period, implementing international guidelines. The nutritional policy of our unit is to achieve full enteral feeding with mother's milk or banked milk from donors as early as possible, and, thus, the preference is to rely on early use of human milk rather than formulae for premature neonates. At our institution, mothers are encouraged to begin expressing breast milk from the time of delivery, and the presence of an efficient Human Milk Bank facility warrants daily availability of pooled human milk from donors if a mother's milk is not available. Accordingly, all neonates who were studied started receiving oral feeding with human milk (either donor milk or mother's milk) via an orogastric or nasogastric tube within a few hours after delivery or, in any case, as soon as the clinical conditions of the neonates allowed such feeding. Enteral nutrition was started with 1 mL of human milk given every 2 or 3 h, and the amount of milk given was increased by 1.0 mL every 3–6 h, as tolerated; human milk was supplemented with parenteral glucose administered from day 1 of life and with amino acids and lipids administered from day 2 of life. Parenteral nutrition initially was provided through an umbilical venous catheter and then was provided through a central venous catheter beginning from the second to fifth days of life and continuing until the end of the need for parenteral nutrition. Nutrition administered via the oral access with intermittent meals or continuous pump feeding was progressively increased, if tolerated, and, concomitantly, parenteral nutrition was progressively decreased and stopped. There was no documentation of errors in the administration of oral or parenteral nutrition, drugs, or infusions, or in the composition of the parenteral nutrition infusions.
Sample collection and isolation and identification of fungi in cultures. Stool samples and gastric aspirate specimens were collected in sterile containers. Oropharyngeal and rectal specimens were collected on swabs (Labobasi).
For the identification of fungi, all specimens were inoculated onto chromogen culture plates (Albicans ID; bioMérieux), which allow for rapid identification of Candida albicans through the blue staining of the colonies after 48 h of incubation at 37°C. Differently stained colonies were speciated through a miniaturized system of biochemical tests (Vitec Yeast; bioMérieux).
Outcome measures and statistical analysis of data. The primary efficacy end point was the overall incidence of enteric fungal colonization. The secondary end points were the intensity of fungal colonization, the ratio of non-albicans versus albicans Candida species in the 2 groups, the incidence of bacterial sepsis (caused by both gram-positive and gram-negative microrganisms), the incidence of sepsis due to LGG, the incidences of both surgical (stage ⩾2, according to Bell's classification [34]) and stage III necrotizing enterocolitis, the incidence of IFI, and the presence of threshold retinopathy of prematurity requiring laser ablation. The variables examined included the total and per-neonate numbers of positive culture results, the incidence of colonization during the first 6 weeks of a stay in the NICU (calculated for all patients and separately for neonates with birth weights of <1000 g and 1001–1500 g in each group), the mean number of fungal isolates obtained from each patient, the incidence of low- and high-grade colonization, the ratio of low-grade colonization to high-grade colonization, the distribution of fungal species (ratio of non-albicans to albicans Candida species in the 2 groups), and the incidence of sepsis due to nonfungal agents (overall and separately for gram-positive organisms, gram-negative organisms, and LGG alone).
The sample size was estimated to be 41 neonates per study arm, and this number was based on the following 2 assumptions: (1) a 2-sided type I error rate of ⩽0.05 for the study, and (2) 80% power to detect an absolute difference of at least 55% in the cumulative incidence of enteric fungal colonization between the untreated and treated groups, given a pretrial incidence of enteric fungal colonization of 52%. SPSS statistical software for Windows, version 8.0 (SPSS Inc.), was used for all statistical computations (Yates' corrected χ2 test or Fisher's exact test was used, as appropriate, to compare proportions and relative risk or risk ratio estimates; differences in risk were used to compare incidence rates among groups, and Student's t test was used for continuous variables). Power calculations were performed using S plus software, version 2000 (Insightful). Analysis of dichotomous outcomes and interpretation of results were performed as suggested in Cochrane Reviewers' Handbook 4.2.2 [35].
Results
Eighty-seven very low birth weight neonates, all of whom were born at our institution, were admitted to our NICU during the study period. Seven were excluded because of the presence of colonization at baseline. Therefore, 80 neonates were thus enrolled and randomized to groups A (n = 39) and B (n = 41); the demographic and clinical characteristics of these 80 neonates are presented in table 1. Between the 2 groups, there were no significant differences in demographic characteristics at baseline, the presence of major risk factors for fungal colonization and IFI [1–10], and the clinical parameters listed in table 1.
Demographic and clinical characteristics of the neonates studied.
Demographic and clinical characteristics of the neonates studied.
The main results of the study are summarized in table 2. The mean number of culture specimens collected from each neonate was similar (P = .38), but the total number of positive culture results, the mean number of fungal isolates recovered from each patient, and the mean number of fungal isolates recovered from each colonized patient were significantly lower in group A (P = .009, P = .005, and P = .005, respectively).
Results and main outcomes.
Results and main outcomes.
The overall rate of gastrointestinal colonization was 23.1% (9 of 39 neonates) in group A and 48.8% (20 of 41 neonates) in group B (P = .01). Cluster analysis of data on birth weight range, the reduction in overall gastrointestinal colonization was significant in neonates with a birth weight of 1001–1500 g (from 42.9% to 14.8%; P = .02), but it was not significant in neonates with a birth weight of <1000 g (P = .27). The distribution of fungal species (the ratio of albicans vs. non-albicans Candida species) was similar, with C. albicans accounting for 78% of the isolates recovered in group A and 70% of those recovered in group B (P = .32).
LGG was never isolated from blood cultures or other specimens, nor was the suspected causal agent of microbiologically undocumented sepsis in any neonate studied. No adverse effects occurred. The incidence of IFI before discharge was slightly lower in group A than in group B (10.3% vs. 12.2%; P = .53), and so was the incidence of surgical necrotizing enterocolitis (0% vs. 2.5%; P = .51).
Discussion
The presence of fungal colonization at various sites of the enteric tube is a well-known risk factor for subsequent fungal dissemination and invasion in preterm neonates [4, 7–9, 36, 37]. Prevention of colonization may be of help in reducing the frequency of development of IFI due to Candida species as well.
Premature neonates in the NICU are highly prone to develop disorders of gut microecology with an overgrowth of pathogenic microflora, including fungi [8], because they are often treated with antibiotics for long courses, and because they experience difficulties in receiving enteral nutrition [10, 38]. The gut is considered to be a major reservoir and an important colonization site for all types of pathogens [6], and it is probably the site from where fungal dissemination usually starts [15].
The present study examined the effectiveness of LGG in preventing gastrointestinal colonization by Candida species in very low birth weight neonates. The results show a significant (2-fold) reduction in gastrointestinal colonization by Candida species among very low birth weight neonates given LGG. The potential mechanisms by which LGG modifies fungal ecology in the gut include a competitive exclusion of fungi, as well as a reduction in their ability to colonize the enteric mucosae via enhanced mucosal IgA responses [39, 40]; changes in intestinal permeability [40, 41], with an increased gut mucosal barrier to fungi [25, 42]; and modifications of the host response to fungal products [31, 32, 43, 44]. Obviously, the results of our study are not attributable to the type of feeding (human milk might positively influence the gut microflora), because the feeding of human milk was the main nutritional policy in both randomized groups. The relative frequencies of Candida subspecies were unmodified in the 2 groups, which means that the beneficial effect of LGG was unspecific (i.e., LGG was evenly distributed over all fungal colonies).
Wainer et al. [18] described a 2-fold reduction in gut colonization by fungi that occurred in association with the use of a topical antifungal drug (miconazole oral gel) for high-risk preterm neonates. Sims et al. [17] administered oral nystatin to preterm neonates with a birth weight of <1250 g, and they found a 4-fold decrease in overall colonization.
Using prophylactic intravenous fluconazole, both Kicklighter et al. [15] and Kaufman et al. [16] achieved a 3-fold reduction in stool or rectal fungal colonization in neonates with a birth weight of <1250 or <1000 g, respectively. In our institution, a 6-year retrospective study that assessed the effectiveness of fluconazole showed a similar (3-fold) reduction in overall colonization in neonates with a birth weight of <1000 g, but it showed a <2-fold reduction in overall colonization in neonates with a birth weight of 1001–1500 g [45]. Anyhow, the promising results obtained with antifungal drugs—notably, with fluconazole—also have to be weighed against the increased risks of selecting resistant strains and against the occurrence of possible changes in drug sensitivities of susceptible strains. In addition, widespread use of prophylactic antifungal agents raises concerns about its cost [20], also because it is currently undefined which subgroups of preterm neonates should be candidates for prophylaxis. Possibly, a combination of strategies could limit these concerns. In the present study, neonates with a birth weight of 1001–1500 g benefited more from the use of probiotics than did neonates with a birth weight of <1000 g, and visual inspection suggests that the positive effect of LGG wanes in neonates with a birth weight of <900 g. The paradox that a greater per-kilogram dosage has a lesser effect is possibly the result of the even greater increase in the overall risk of colonization as birth weight decreases. Higher LGG dosages might have enhanced the direct positive effect that the probiotic has on immunity in neonates who received lesser benefit; however, the per-kilogram dosages used are considerably higher than those recommended by the producing company, and the possibility that the probiotic will be carried indefinitely raises a safety concern. Larger dose-finding studies could be performed to better address this issue.
Combinations of drugs and probiotics might be of beneficial use for all very low birth weight neonates, with greater reliance on probiotics occurring in larger neonates and lesser reliance occurring in smaller neonates. Such combinations should allow a reduction in the drug dosage needed to achieve suppression of colonization. Clearly, in each case, the optimal combination should not be a strict reflection of the current separation between neonates with a birth weight of <1000 g and neonates with a birth weight of 1001–1500 g, and they might be adjusted as required. For example, more reliance could be placed on the use of antifungal drugs in NICUs with higher frequencies of colonization and invasive fungal disease.
On the basis of both current drug prices in Italy and recommended fluconazole schedules for premature neonates [15, 16], 1 month of prophylaxis for a neonate with a birth weight of 1001–1500 g would cost €25 with fluconazole, but it would cost only €15 with LGG. Furthermore, 3 of 4 very low birth weight neonates who are admitted to NICUs weigh 1001–1500 g at birth, and, for this group, the incidence of fungal colonization and subsequent progression to IFI is lower than that for neonates who weigh <1000 g at birth. If prevention of fungal colonization were to rely more on LGG than on fluconazole for neonates with a birth weight of 1001–1500 g, this would allow relevant cost savings for the subgroup of preterm neonates in whom IFI is less frequent and less severe (i.e., those for whom prophylaxis has a relatively unfavorable cost-effectiveness ratio).
To date, only a few clinical trials have reported the outcomes for preterm neonates given probiotics. In our series, a significant reduction in fungal colonization was achieved with no associated adverse effect—in particular, with no onset of sepsis caused by LGG. Although our findings regarding colonization are encouraging, the hypothesis that LGG or other probiotics may be effective in the prevention of IFI will have to be tested in larger, more definitive, ad hoc trials.
Acknowledgments
Potential conflicts of interest. All authors: no conflicts.
Comments