Introduction

Bloodstream infections are major causes of morbidity and mortality and represent the eighth leading cause of infant death [1]. Candidemia is the second most common nosocomial bloodstream infection in very low weight neonates and children [2, 3]. Mortality rates associated with candidemia remain significant, despite use of advanced therapeutic approaches [3]. Although most data collected are from adults, available evidence indicates that Candida species cause a high proportion of bloodstream infections among children, especially neonates [4]. Neonatal candidemia frequently complicates clinical course of preterm neonates, especially those with underlying disease [5], whereas, central nervous system involvement and neurologic impairment is common. Central venous catheters and arterial lines, parenteral nutrition, mechanical ventilation and extended use of antimicrobials enhance the risk of neonatal candidemia [6]. The aim of this study was to investigate the prevalence of candidemia and etiological agents, predisposing factors and outcome among neonates, during a 5-year period. Furthermore, a review of recent published case series is presented.

Materials and Methods

A retrospective cohort study of bloodstream infections caused by Candida species was carried out in Neonatal Intensive Care Unit (NICU) at the University General hospital of Patras, a level ΙΙΙ referral unit, located in southwestern Greece. The study took place over a 5-year period (2005–2009). NICU consists of 20 beds with about 350 admissions per year. All candidemia cases were evaluated in terms of mycological data including species isolation and antifungal susceptibility, as well as, patient data including demographic data, risk factors, and outcome. Candida isolation from blood cultures was performed by BacT/Alert 3D (Biomerieux SA, Marcy l’Etoile, France). All isolates were identified using the germ tubes test and API 20C AUX system (Biomerieux SA, Marcy l’Etoile, France). Antifungal susceptibility was carried out by E test (AB Biodisk) on RPMI–2% glucose agar and MICs for amphotericin B, 5-flucytosine, ketoconazole, itraconazole, fluconazole, voriconazole, posaconazole, and caspofungin were evaluated according to the manufacturer’s instructions, as previously described [7]. Resistance to 5-flucytosine was defined as MIC ≥32 mg/L, to fluconazole ≥64 mg/L, to itraconazole ≥1 mg/L, to voriconazole ≥4 mg/L. MIC cutoff value for susceptibility to caspofungin and amphotericin B was 2 and 1 mg/L, respectively [8–10]. Isolates demonstrating MIC ≤1 mg/L were considered susceptible to posaconazole [8, 11], while strains with MIC ≥1 mg/L were interpretive as resistant to ketoconazole.

All infants nursed were included in the study. A candidemia case was defined as a positive blood culture with a Candida spp in an eligible subject during the study period [12]. Persistent Candida bloodstream infection has been defined in various studies as candidemia with duration greater than 2–5 days in spite of adequate dosing of conventional antifungals [13]. Infants with birth weight less than 2,500 g were defined as low birth weight, less than 1,500 g of very low birth weight and less than 1,000 g were considered as extremely low birth weight infants (ELBW) [13]. Prematurity was defined as gestational age of less than 37 weeks [14]. Other risk factors included presence of central venous catheters, endotracheal intubation, total parenteral nutrition, previous use of antibiotics, steroids, duration of stay in NICU, previous bacteremia, respiratory disorders, jaundice, liver and renal deficiency, central nervous system disorders, congenital malformations involving heart and large vessels, congenital malformations involving the alimentary tract, complications involving the intestinal tract (bowel perforation, necrotizing enterocolitis). Statistical analysis was performed using SPSS for Windows v.17.0.1 [15]. Differences in proportions were evaluated using the Chi-square test. Statistical significance was defined as P < 0.05.

Results

Mycological Data

During the study period, forty candidemia cases were treated in NICU. The species identified were in twenty-seven cases C. albicans (67.5%), in ten cases C. parapsilosis (25%), in two cases C. tropicalis, and in one case C. glabrata. A decreasing pattern in candidemia incidence was observed during the study period, from eleven cases in 2005 and twelve in 2006 to seven cases (2007), and five cases during the last 2 years of the survey. In two patients, Candida was isolated from cerebrospinal fluid, and in four cases, echocardiograms were consistent with endocarditis [16]. No cases of Candida endophthalmitis/retinitis, isolation of Candida spp from a sterile cavity, lungs or other organs was observed. However, Candida was also isolated from urine in five candidemia cases.

All C. parapsilosis and C. tropicalis strains were susceptible to amphotericin B, 5-flucytosine, azoles, and caspofungin (Fig. 1). All C. albicans strains were susceptible to amphotericin B and 5-flucytosine, and only one strain was resistant to caspofungin (susceptibility rate 96.5%). In terms of susceptibility to azoles, voriconazole seems to be the most efficient (93%), whereas, susceptibilities to fluconazole, itraconazole, ketoconazole, and posaconazole were 89, 81.5, 85, and 87.5%, respectively (Fig. 1).

Fig. 1
figure 1

Antifungal susceptibility rates. AM amphotericin B, FC 5-flucytosine, FL fluconazole, KE ketoconazole, IT itraconazole, VO voriconazole, PO posaconazole, CS caspofungin

Full size image

Demographic Characteristics and Risk Factors

Thirty-four candidemia cases (85%) occurred at low birth weight neonates, and thirty-three cases (82.5%) at low gestational age neonates. Previous use of antibiotics and steroids was reported in 36 cases (90%). Invasive procedures were common at NICU patients, such as use of central indwelling catheters (75%), endotracheal intubation (80%), and total parenteral nutrition (95%). All neonates with candidemia had at least one episode of bacteremia before its occurrence. Gastrointestinal pathology such as necrotizing enterocolitis was identified in 12.5% of candidemia patients. Respiratory disorders were observed in 19 neonates with candidemia (47.5%), jaundice in 10 (25%), CNS disorders in 4 (10%), heart and vessels malformations in 6 (15%), and alimentary tract malformations in 3 (7.5%) neonates. Liver and renal failure was observed also in 3 neonates (7.5%).

Evaluation of Isolated Candida spp and Specific Risk Factors

Isolation of C. albicans or non-albicans Candida and presence of specific risk factors was evaluated. Statistical data revealed that low birth weight is associated with candidemia due to C. albicans (P < 0.05). C. albicans was isolated at a higher rate, eleven among twelve candidemia cases involving ELWB (P < 0.01). No statistically significant differences were observed between C. albicans and Candida non-albicans candidemia regarding presence of other risk factors (Table 1).

Table 1 Risk factors for neonatal candidemia
Full size table

Outcome

Despite appropriate diagnosis and treatment, fifteen neonates with candidemia died (35.7%,). Fourteen had low birth weight (less than 2,500 g), eleven less than 1,500 g, and nine less than 1,000 g. Low birth weight and low gestational age were associated with poor outcome (P < 0.01). Presence of renal failure was also associated with poor outcome (P < 0.05). No significant association between outcome and presence of endotracheal intubation, central lines, parenteral alimentation or other risk factors was found (P > 0.05) (Table 2). In addition, no statistically significant differences in survival rates between neonates with candidemia caused by C. albicans (63%) and neonates with candidemia caused by Candida non-albicans (61.5%) were observed (Table 1). During the study period, the only one candidemia case due to C. glabrata involved an ELWB neonate and was fatal, whereas, both two cases due to C. tropicalis were efficiently treated.

Table 2 Risk factors and neonatal outcome
Full size table

Cases of Persistent Candidemia

Thirteen candidemia cases (32.5%) were identified as persistent candidemia (duration more than 7 days–five cases more than 14 days), three of them were complicated with endocarditis, whereas, three (23%) were fatal. Seven persistent candidemia cases (54%) were due to C. albicans and six (46%) due to C. parapsilosis. Eight neonates with persistent candidemia were less than 1,500 g and ten less than 2,500 g. In three cases, Candida spp was isolated from urine.

Discussion

There are several important observations in this study. Review of published case series on epidemiology, risk factors, and outcome of NICU candidemia is summarized in Table 3. C albicans is also the most common blood isolate in our NICU (67.5%) [17–28]. Similarly, C. parapsilosis is the second commonest isolate and constitutes one-fourth of blood isolates (25%) [17–22, 25–27]. This finding may be attributed to the use of central catheters [29], parenteral alimentation at premature infants [5], gastrointestinal colonization [5], and transmission through the hands of health care workers [28]. In some areas, mainly Australia, South America/Brazil, Spain, and Portugal, C. parapsilosis counts for the majority of candidemia cases [5, 12, 30–34]. C. glabrata and C. tropicalis are uncommon in neonates (2.5 and 5%, respectively), whereas, less common Candida species, such as C. krusei, C. lusitaniae, and C. guilliermondii, are absent.

Table 3 Review of published case series
Full size table

According to current guidelines, both amphotericin B and fluconazole appear acceptable choices for the therapy of neonatal candidemia [35]. Candida isolates from neonatal candidemia are susceptible to most agents [5, 12, 18, 25, 27, 34]. More specifically, amphotericin B has an excellent antifungal activity, whereas, fluconazole is active against all C. parapsilosis isolates. It exhibits an antifungal action to 89% of C. albicans isolates. Voriconazole seems a very promising agent exhibiting the highest susceptibility rate (93%) among azoles. It has been shown that voriconazole has good distribution into brain, lungs, kidneys, liver, and spleen and demonstrates fungostatic activity against Candida spp [6]. Nevertheless, in the clinical setting, precaution should be taken to ensure adjusting dosing and prompt identification of adverse events, even rare, such as visual effects and liver toxicity [36]. Also, caspofungin demonstrates high in vitro antifungal activity, although in clinical use it has some disadvantages due to poor penetration through uninflamed meninges and ocular vitreous [6]. In our NICU, use of lipid formulations of amphotericin B is more favored for candidemia treatment, whereas, recently, caspofungin is used in selected cases. In a recent survey that took place in Greece, amphotericin B was the drug of choice for treating neonatal candidiasis [37]. As shown on Table 3, amphotericin B and to a lesser extent fluconazole are preferred for neonatal candidemia treatment [5, 19, 22, 23, 25–27, 34].

During the study period, thirteen persistent candidemia cases (32.5%) were identified, although the in vitro resistance to antifungal drugs was very low. Persistence of infection despite treatment with an appropriate dosage of an active agent is well recognized. It is attributed to neonate’s immune status that is unable to eliminate the fungus [13]. In previous studies, persistent candidemia was observed in 10% of ELBW infants [12, 13], was associated with non-albicans infection [38] and exhibited a poor outcome [13]. In our setting, there is no difference in Candida spp isolates (54% due to C. albicans and 46% due to C. parapsilosis), whereas, the mortality rate is not higher that overall mortality (23 vs. 35.7%).

Risk factors associated with candidemia in NICU have been well characterized [6, 13, 39] and were common in the present study. Almost all patients had at least one identifiable risk factor, such as, low birth weight (85%), low gestational age (82.5%), previous use of antibiotics and steroids (90%), use of central indwelling catheters (75%), endotracheal intubation (80%), total parenteral nutrition (95%), previous bacteremia (100%), gastrointestinal pathology such as necrotizing enterocolitis (12.5%), respiratory disorders (47.5%), jaundice (25%), CNS disorders (10%), heart and vessels malformations (15%), malformations of alimentary tract (7.5%), as well as, liver and renal failure (7.5%). As shown on Table 3, most studies focus on birth weight, gestational age, use of antibiotics and steroids, parenteral nutrition, presence of central venous catheters, endotracheal intubation, abdominal surgery, gastrointestinal tract disorders or colonization and length of ICU stay. Vaginal delivery as a risk factor is controversial [18, 20].

C. albicans is isolated at a higher rate in ELWB (P < 0.01) in our NICU, whereas in a large scale study, neonates with C. parapsilosis bloodstream infection were older than neonates with C. albicans infection (P < 0.001), [17]. In accordance with other studies, no statistically significant differences were observed between C. albicans and Candida non-albicans candidemia regarding presence of other risk factors [27].

Although low number of patients may restrict parameters with statistically significant results, low birth weight and low gestational age constitute significant determinants of candidemia outcome. The Expert Panel recommends routine fluconazole prophylaxis for premature infants and infants with extremely low birth weights in nurseries that have a high incidence of invasive candidiasis [35]. No prophylaxis is given to premature infants and infants with ELBW by our NICU personnel. Recent studies suggest that fluconazole prophylaxis may exert a beneficial effect on extremely preterm neonates even in nurseries with low candidemia incidence rates [40]. Also, emergence of natively fluconazole-resistant Candida spp should be taken into account, although existing data do not justify this theory [41]. Our data show no significant association between outcome and presence of endotracheal intubation, central lines, parenteral alimentation or other risk factors (P > 0.05). In previous reports, duration of hospitalization [24], surgery [27], endotracheal intubation [22], failure [33] or delay [19] to remove a catheter and lack of antifungal therapy [33] were associated with higher mortality rates in pediatric patients with candidemia. Although failure to remove catheters was an independent risk factor for early death, it had no impact on late death group (8–30 days after candidemia was diagnosed) [33].

In our setting, overall mortality is 35.7%, whereas in other studies, it varies from 10.2 [26] to 54.2% [24]. In accordance with other studies [22], no statistically significant difference is observed regarding outcome of C. albicans versus outcome of Candida non-albicans candidemia. Nevertheless, C. albicans association with higher mortality rates in neonates has been described [14, 27]. Interestingly, in our setting, C. albicans is isolated at a higher rate, eleven among twelve candidemia cases involving ELWB (P < 0.01), exhibiting high mortality rate (73%). In the ELBW group, no case is attributed to C. parapsilosis candidemia. In contrast, in the case of neonates more than 1,000 g, C. parapsilosis is associated with higher mortality rate (40%) than C. albicans (12.5%), (P > 0.05).

In conclusion, C. albicans represents the most common Candida (67.5%), isolated from candidemia cases in our NICU, followed by C. parapsilosis (25%). Resistance to antifungal agents is rare and involves mainly azoles. C. albicans is most often isolated from ELBW neonates as compared to non-albicans Candida (P < 0.01). ELBW and low gestational age are associated with poor outcome (P < 0.01). Taking into account the high mortality rates of neonatal candidemia and the high cost of hospital care [42], therapeutic approach should be multifaceted. It should include prevention by improving infection control practices and antimicrobial stewardship strategies [43], identification, and if possible, elimination of predisposing factors, as well as, early diagnosis and prompt appropriate treatment according to current guidelines and epidemiologic trends.