Invasive candidiasis in a Brazilian neonatal intensive care unit

,


Introduction
Invasive candidiasis is an important cause of morbidity and mortality among neonates admitted to the Neonatal Intensive Care Unit (NICU). Risk factors associated with invasive candidiasis include prematurity, very low birth weight, vascular catheters, parenteral nutrition, administration of broad-spectrum antibiotics, abdominal surgery, prolonged hospitalization, and artificial ventilation. [1][2][3] Incidence is especially high among extreme low birth weight (ELBW) neonates (those born with <1000g), and it has ranged from 2% to 15%. [2][3][4][5] Therefore, different strategies have been developed to prevent and improve outcomes related to invasive candidiasis among ELBW neonates. The two main strategies are 1) empiric antifungal therapy, which has reduced Candida associated mortality, [6][7][8] and 2) fluconazole prophylaxis, which has had an impact in reducing the incidence of invasive candidiasis with no adverse outcomes as fluconazole toxicity or emergence of resistant Candida species. 9,10 Nonetheless, the majority of evidence comes from developed countries; there is scarce data about candida epidemiology in the neonatal population in developing countries, especially from Latin America, where the adoption of fluconazole prophylaxis is not standard practice. Moreover, no attention has been given to candidiasis for neonates with birthweight ≥1000g with risk factors as abdominal surgery. The reported incidence in neonates with birth weight >1500g is below 1%. 11 Therefore, we decided to study invasive candidiasis from a tertiary NICU, a referral center for neonatal surgery and preterm birth, located in Brasilia, Brazil, during a period that fluconazole prophylaxis was not implemented, comparing ELBW neonates and neonates with birthweight ≥1000g, in order to support our decision to begin fluconazole prophylaxis or not.

Methods
We performed a retrospective, cross-sectional study of all cases of invasive candidiasis identified in our service from April 2015 to June 2018, comparing ELBW neonates with those with birth weight ≥1000g. Our hospital is the largest public maternity hospital in Brasilia, Brazil. It is a regional referral center for preterm-births and neonatal surgery; it has a 30-bed NICU and an intermediate care unit with 16 beds.
We obtained a list of Candida species isolated in blood or cerebral spinal fluid from the microbiology laboratory database and reviewed the patients' records. We collected the following data: gestational age, birthweight, sex, type of delivery, history of abdominal surgery, date of the first positive culture for Candida, platelet count in the occasion of suspected fungal infection, the duration in days of use of mechanical ventilation (MV), central venous catheter (CVC), total parental nutrition (TPN) and antibiotics prior of suspected fungal infection, use of antifungal therapy, duration of hospital stay and patients' outcome: death or hospital discharge.
Neonates with clinical suspicion of infection had their blood samples collected in the volume of 1 mL and immediately inoculated in bottles for blood culture. Samples were initially submitted to a continuous automated system for microorganism detection (Bactec fluorescent series system ® Becton Dickinson Microbiology System). Then, a gram stain was performed and an automated panel of fungal identification (MicroScan Walk-Away ® Dade Behring Inc.) was used to identify fungal species.
We analyzed the data descriptively, using proportions for discrete variables and median and inter quartile range (IQR) for continuous variables. We used Chi-square test, or Fisher test when appropriate, for discrete variables and Kruskal-Wallis test for continuous variables to compare ELBW neonates with neonates with birth weight ≥1000g, considering a p value <0.05 as statistically significant.
The study was approved by the Ethics and Research Committee of Fundação de Ensino e Pesquisa em Ciências da Saúde -FEPECS (Foundation of Teaching and Research in Health Sciences), CAAE: 81693317.6.0000.5553, under the number 2.514.300.
Considering the 38 cases of invasive candidiasis, the median gestational age was 33 weeks (IQR 28 -37) and the median birth weight was 1405g (IQR 900-2265). Nineteen (50%) were male, 24 (63%) were born through cesarean delivery and 24 (63%) had a previous abdominal surgery. The median age during invasive candidiasis onset was 20.5 days old Silva HYW, Freitas FTM Invasive neonatal candidiasis in Brazil (IQR 13 -37). In seven cases (18%), invasive candidiasis was the first episode of late-onset sepsis. All cases presented low count of platelets on the day of onset of symptoms and all cases had a previous use of antibiotics; the median duration of antibiotic use was 14.5 days (IQR 9 -25). Thirty-one cases (82%) had previous use of broad-spectrum antibiotic (ampicillin-sulbactam, piperacilin-tazobactam, cefotaxime, cefepime, meropenem or vancomycin). All cases received TPN; the median duration of TPN use was 17.5 days (IQR 3 -34). All cases had a CVC, the median duration of CVC use was 19 days (IQR 12 -36). Thirty-five cases (92%) received MV and the median duration of MV was 7.5 days (IQR 4 -20). The median duration of NICU stay was 52.5 days (IQR 36 -108). Three cases (8%) died before receiving any antifungal therapy, 30 (79%) received antifungal therapy with amphotericin B and 5 (13%) with fluconazole. There were 12 deaths and a fatality rate of 32%. Table 1 summarizes the characteristics between neonates of ELBW and those with birth weight ≥1000g. Previous abdominal surgery (85% vs. 17%; p <0.01) and days of antibiotic use prior to Candida infection (18 vs. 10.5; p = 0.04) were statistically more frequent among neonates ≥1000g of birth weight. On the other hand, a longer period in days of MV (10 vs. 5.5; p = 0.04) was statistically more frequent among ELBW neonates.
It was not possible to identify the Candida species in 10 (26%) cases. From the 28 remaining samples with species identification, 18 (64%) were Candida non-albicans, although C. albicans was the main specie isolated (36%) ( Table 2).

Discussion
Our data shows that invasive candidiasis is a relevant problem in our NICU. Mortality is high, we found a fatality rate of 32%, similar to the 34% found in a prospective cohort of ELBW neonates, which took place over three years at 19 centers in the United States. 1 The observed incidence is similar to that reported elsewhere, varying from 0.5% to 2% overall and 5% to 15% among ELBW. [2][3][4][5] Data from other Brazilian unit, located in Belo Horizonte, found a higher incidence among ELBW neonates of 9.35% and 5.25% when considering the whole cohort of very low birth weight neonates (< 1500g) during a three-year period, applying different strategies of fluconazole prophylaxis. 12 Nonetheless, our incidence of 2% among neonates with birth weight ≥ 1000g is especially high compared to a multicenter study that found an incidence of 0.06% among neonates with birth weight > 1500g. 11 This may be explained by the high proportion of abdominal surgery seen among neonates with birth weight ≥1000g with invasive candidiasis in our study. We are a referral center for neonatal surgery and receive a large number of neonates with abdominal malformations and enterocolitis for surgery. Due to abdominal surgery, these neonates are exposed to prolonged time of TPN and broad-spectrum antibiotics, accumulating different risk factors for invasive candidiasis. In fact, gastrointestinal pathologies caused by major malformations were reported as an important risk factor for Candida infections among heavier neonates. 11,13,14 Thus, a large number of neonates with invasive candidiasis who underwent abdominal surgery may not receive prophylactic fluconazole, as it is usually prescribed for neonates <1000g. This was already observed in a previous study, which reported that 58% of neonates with invasive candidiasis did not qualify for fluconazole prophylaxis. 14 Additionally, the exposure to potentially modifiable risk factors, such as use of invasive devices, TPN and antibiotics, was long. Invasive devices, such as CVC may become colonized with Candida and MV may promote colonization of the respiratory and gastrointestinal tract by not allowing normal mucociliary clearance with an endotracheal tube. The use of intralipid emulsions and TPN may have a direct causal effect such as enhancing the growth of fungal pathogens or this may be confounded by delayed enteral feedings. Additionally, it has been shown that the use of broad-spectrum antibiotics may disrupt normal intestinal flora and is associated with invasive candidiasis in different NICUs. 15 This is important because these risk factors may be modified by a clinical care capable to prevent Candida infections.
Candida albicans was the main specie isolated, although the majority of species were non-albicans Candida. As observed in more recent studies, while C. albicans is still the main specie, there is an increasing number of non-albicans Candida causing infections in neonates. 16 Data from 23 units in 8 countries from Latin America, including Brazil, found C. albicans as the main specie (44%), but nonalbicans Candida predominated (56%), being C. parapsilosis and C. glabrata the most isolated, with a low proportion of fluconazole resistant species. 17 It is noteworthy to mention that we isolated only a single fluconazole resistant specie, one case of C. glabrata and none of C. krusei. As we did not use fluconazole prophylaxis and our first option to treat Candida infections was amphotericin B, there was prescription practices, among other factors that may differ widely among centers. Nevertheless, its strength is to picture the epidemiology of invasive candidiasis in a highly complex Brazilian NICU before fluconazole prophylaxis, as such strategy is not standard practice in the country. Furthermore, it gives a special attention to heavier neonates that performed abdominal surgery.
In conclusion, after this analysis, we decided to start fluconazole prophylaxis for ELBW because our fatality rate was high and our incidence approached 5% as it is usually recommended. 19,20 Fluconazole prophylaxis for ELBW has proved to be effective in reducing Candida invasive infection incidence and safe to not provide relevant adverse reaction or development of Candida resistant strains. 9,10 Future analytic studies, with a prospective design and groups of comparison, including multiple centers and enough sample are necessary to investigate no inducible pressure on fluconazole intrinsic resistant Candida strains as it might have had on amphotericin B resistant strains, such as C. lusitaniae and C. guilliermondii. The proportion of C. lusitaniae was higher than observed elsewhere; 17 some investigators have theorized that the widespread use of amphotericin B empiric antifungal therapy selects for infections with C. lusitaniae and C. guilliermondii. 18 Our study has limitations; this is a relative small retrospective study with no purpose to evaluate risk factors or the effectiveness of prevention measures. Because of laboratory limitations, we were not able to identify Candida species in a quarter of samples. Moreover, the study represents a single unit, with no possible extrapolation for a broader population, especially when it comes to Candida infection that depends of many factors as the proportion of preterm neonates admitted, neonatal surgery and antibiotic  fluconazole prophylaxis for neonates with birth weight ≥1000g who have gastrointestinal pathologies and undergo abdominal surgery, since this group may consist in a high-risk population for Candida invasive infection in highly complex NICU.

Author's contribution
Silva HYW and Freitas FTM contributed in the concept of the article and did the data analysis. Silva HYW collected the data and wrote the first draft. Freitas FTM reviewed the first draft and wrote the final version. All authors approved the final version of the article.