Candida spp. Isolation from Critically Ill Patients’ Respiratory Tract. Does Antifungal Treatment Affect Survival?

Abstract

Introduction: Isolation of Candida spp. from bronchial samples of patients on mechanical ventilation is common. Even though it may not always reflect infection, it may induce immunological changes that can facilitate bacterial pneumonia. In this case, antifungal treatment is of uncertain value. This study examined the impact of antifungal treatment on the outcome of intensive care unit (ICU)-acquired respiratory tract infection (RTI) of critically ill, immunocompetent patients, with Candida isolation from their respiratory tract. Methods: This is a retrospective cohort study of adult patients hospitalized in the ICU of the University Hospital of Heraklion, Greece, from 2014 through 2016 with ICU-acquired RTI and Candida spp. isolated from their bronchial secretions. Data regarding medical history, demographics (gender, age), reason for ICU admission, previous antimicrobial use or hospitalization, SOFA and APACHE II score, clinical outcomes (primary clinical outcome: overall mortality during hospitalization; secondary clinical outcome: mortality during the ICU stay and duration of ICU and hospital stay) at the end of their ICU stay and at the end of their hospital stay were recorded and consequently evaluated. A logistic regression analysis model evaluated the effect of the recorded parameters in association with ICU mortality and overall mortality during hospitalization. Results: A total of 90 individuals were enrolled. Of them, 47 (52.2%) were treated with antifungals during their hospitalization around the time of Candida isolation. Patients treated with antifungals had higher SOFA and APACHE II scores, longer duration of stay in the ICU, more days on ventilator and higher total mortality during hospitalization. Multivariate logistic regression analysis identified antifungal use to be independently associated with total mortality during hospitalization. Conclusions: Antifungal use in patients with ICU-acquired RTI was associated with higher overall mortality as compared to those not receiving such agents.

Introduction

Isolation of Candida spp. from the bronchial samples of patients on mechanical ventilation (MV) is common. It is estimated to occur in about 30% of patients on MV for more than 2 days, and in about 50% of those suspected to have a ventilator-associated pneumonia (VAP) [1]. Isolation of Candida spp. from respiratory tract samples is associated with increased duration of MV, increased intensive care unit (ICU) and hospital stay, as well as worse outcomes [2,3,4]. Furthermore, Candida colonization of the respiratory tract may facilitate subsequent pneumonia, as shown in a study using a rat model colonized in the respiratory tract by C. albicans and subsequently infected by Acinetobacter baumannii [5]. In that study, rats with their respiratory tract colonized by C. albicans developed experimental pneumonia more often, had higher bacterial loads and had lower interleukin 17 (IL-17) and higher gamma interferon (IFN-γ) concentrations in the lungs compared to controls [5]. Another study showed that rats’ respiratory tract colonization by C. albicans elicited a Th1-Th17 immune response, favoring the development of bacterial pneumonia by inhibiting bacterial phagocytosis by alveolar macrophages, while antifungal treatment decreased the risk of bacterial pneumonia [6]. In another rat model, C. albicans colonization of the respiratory tract has been shown to impede alveolar macrophage reactive oxygen species production and was correlated with increased rates of Pseudomonas aeruginosa pneumonia [7].

Candida pneumonia is rarely diagnosed due to difficulties in distinguishing colonization from infection. However, Candida may cause pneumonia in severely immunocompromised individuals either due to aspiration of Candida spp. colonizing the oral cavity or as part of disseminated candidiasis [8].

There are data suggesting that Candida colonization of the respiratory tract may possibly facilitate the development of pneumonia [9]. Although data are contradictory, co-colonization of the respiratory tract by both bacteria and fungi may augment the risk of infection, facilitate the development of pneumonia and increase the disease’s severity [10]. Candida can produce biofilm, which is a network of hyphae, pseudohyphae and yeasts in a matrix of polysaccharides, proteins and other factors, resulting in the protection of fungi from host defenses and the development of antifungal resistance [11]. Within the biofilm, several interactions between fungi and bacteria through quorum sensing are possible, through a wide range of cell to cell communications and metabolic processes, leading to synergistic or antagonistic interactions [11,12,13,14]. For example, in an experimental model of rats injected with C. albicans, increased production of inflammatory mediators such as IL-6, TNF-α and IFN-γ led to inhibition of phagocytosis by alveolar macrophages leading to changes in airway microecology and increased likelihood of colonization by P. aeruginosa, while similar results also exist regarding increased colonization by S. aureus and Enterobacteriaceae [13,14]. In another example, quorum sensing between C. albicans and A. baumannii leads to changes of morphology of C. albicans and changes in movement and production of virulence factors of A. baumannii, likely resulting in increased alveolar invasiveness of A. baumannii in mice colonized by Candida [5,12]. However, critically ill patients with respiratory tract colonization by Candida receiving antifungal treatment, such as with aerosolized amphotericin B, have not had a lower risk for VAP, despite reduction of Candida colonization [15]. Empirical antifungal treatment of patients with VAP and Candida respiratory colonization did not affect the diseases’ outcome, while treatment with micafungin did not affect the risk of VAP in immunocompetent mechanically ventilated patients [16,17].

Moreover, colonization by Candida as well as antifungal treatment may alter the fungal microbiome of patients, and these alterations, along with further changes of the rest of the microbiome due to concomitant antimicrobial use may negatively affect patients’ survival [18,19,20]. Thus, the effect of antifungal treatment of critically ill patients with Candida isolation from their respiratory secretions, remains uncertain.

The aim of this study was to examine the impact of antifungal treatment on outcome of critically ill immunocompetent patients, with ICU-acquired respiratory tract infection (RTI) and Candida spp. isolation from their respiratory tract.

Methods

Study population

This is a retrospective cohort study of adult patients hospitalized in the ICU of the University Hospital of Heraklion, Greece, from January 2014 through December 2016 with ICU-acquired RTI and Candida spp. isolation in their bronchial secretions. ICU-acquired RTI was defined as a new RTI in a patient hospitalized in the ICU for at least 48 h. More specifically, eligibility criteria were the following: onset of symptoms after 48 h of admission to the ICU and new or evolving infiltrate on chest X-ray or computerized tomography or lung ultrasonography along with auscultatory findings consistent with pneumonia/pulmonary consolidation (e.g., rales, dullness on percussion or bronchial lung sounds) or hypoxemia, new onset of cough or worsening of pre-existent hypoxemia or new onset need for mechanical ventilation along with fever of at least 38 °C or elevated total peripheral white blood cell (WBC) count to more than 10,000/mm³ or leukopenia with a WBC count of less than 4,000/mm³. All previously immunocompetent adult patients hospitalized in the ICU for at least 48 h that developed an ICU-acquired RTI and had Candida spp. isolated in their bronchial secretions [tracheal aspirate, sputum or bronchoalveolar lavage (BAL)] were included. Data regarding demographics (age, gender), medical history, reason for ICU admission, antimicrobial use during the previous month or hospitalization during the last three months, APACHE II and SOFA scores, antifungal treatment for any reason ±10 days around Candida spp. isolation, clinical outcomes at the end of their ICU stay and at the end of their hospital stay were recorded and evaluated. The primary clinical outcome of this study was overall mortality during hospitalization, and the secondary outcome was mortality during the ICU stay and duration of ICU and hospital stay. Patients with previous immunosuppression (neutropenia, bone marrow or solid organ transplantation, corticosteroid treatment consisting of >2 mg/kg of methylprednisolone or equivalent) were excluded.

The study was approved by the Ethics Committee of the University Hospital of Heraklion, Crete.

Microbiological methods

Samples of the lower respiratory tract (sputum, tracheobronchial aspirate and BAL) were collected for quantitative bacterial and fungal cultures. Only sputum and tracheal aspirates of high quality (i.e., with more than 25 leucocytes and fewer than 10 squamous epithelial cells per high-power field) were processed for culture. For the isolation of Candida spp., specimens were inoculated on Sabouraud dextrose agar supplemented with chloramphenicol (50 mg/mL) (Oxoid Limited, UK) and incubated at 36 °C for 72 h. Results were expressed as colony forming units (CFUs) per milliliter (mL) (CFU/mL). For BAL and sputum cultures the accepted threshold was ≥10⁴ CFU/mL and for endotracheal aspirate cultures it was ≥10⁶ CFU/mL. Identification to the species level was performed by sugar assimilation tests using the API 20 C AUX system and/or the Vitek 2 automated system (both BioMérieux, France). Susceptibility to antifungal agents was determined using the Vitek 2 system.

Statistics

Categorical data were analyzed with Fisher’s exact test. Continuous variables were compared using Student’s t-test for normally distributed variables and the Mann–Whitney U-test for non-normally distributed variables. All tests were two-tailed and p-values <0.05 were considered significant. Data are presented as numbers (%) for categorical variables and median [interquartile range (IQR)] or mean [±standard deviation (SD)] for continuous variables. A logistic regression analysis model was developed to evaluate the effect of gender, age, weight, body mass index (BMI), APACHE II and SOFA score on admission, prior hospitalization, prior surgery, antimicrobial use, tracheostomy, positive blood culture(s), antifungal agents used and duration of ICU stay with ICU mortality and overall mortality during hospitalization. All were calculated with GraphPad Prism 6.0 (GraphPad Software, Inc., USA). A multivariate logistic regression analysis model was developed to evaluate the association of factors identified in the univariate analysis with a p<0.1 with overall mortality in the hospital. Multivariate analysis was performed using SPSS version 23.0 (IBM Corp., USA).

Results

In total, 90 individuals, (67 males; 74.4% and 23 females; 25.6%) with a mean age of 67.1±15.9 years were evaluated. The most common patients’ comorbidities were hypertension in 42 (46.7%), chronic pulmonary disease in 25 (27.8%), diabetes in 31 (34.4%), congestive heart failure in 19 (21.1%), neurological disorders in 11 (12.2%), and myocardial infarction in 8 (8.9%). The most common reasons for ICU admission were respiratory failure in 36 patients (40%), trauma in 15 (16.7%), septic shock in 10 (11.1%) and intracranial bleeding in 9 (10%).

Table 1. Characteristics of patients with Candida spp. isolation from the respiratory tract.

	All patients (n=90)	Treated with antifungals (n=47)	Not treated with antifungals (n=43)	p	t and df or U value	Effect size (Cohen’s d)
Age, mean±SD, years	67.1±15.9	68±15.1	66.2±16.8	0.587	t=0.5454 df=88	0.116
Male, n (%)	67 (74.4)	32 (68.1)	35 (81.4)	0.226	NA	NA
Prior hospitalization, n (%)	46 (51.1)	29 (61.7)	17 (39.5)	0.057	NA	NA
Prior antimicrobial use, n (%)	38 (42.2)	23 (48.9)	15 (34.9)	0.205	NA	NA
APACHE II score on admission, median (IQR)	22 (15-27)	23 (19-28.8)	20.5 (15-24)	0.031	U=640.5	0.664
SOFA score on admission, median (IQR)	10 (7-11)	11 (8.3-12)	9 (7-11)	0.042	U=654.5	0.636
Days on ventilator, median (IQR)	15.5 (7-27)	19 (9-30)	11 (6-19)	0.027	U=737.0	0.479
Tracheostomy, n (%)	55 (61.1)	32 (68.1)	23 (53.5)	0.196	NA	NA
Positive blood culture, n (%)	60 (66.7)	31 (66)	29 (67.4)	1	NA	NA
Duration of stay in ICU, days, median (IQR)	18.5 (10.8-30)	21 (13-33)	17 (9-25)	0.012	U=699.5	0.549
Duration of stay in hospital, days, median (IQR)	34 (21-49.5)	38 (23-49)	31.5 (19-50)	0.278	U=854.5	0.268
ICU mortality, n (%)	20 (22.2)	12 (25.5)	8 (18.6)	0.456	NA	NA
Total hospital mortality, n (%)	33 (36.7)	24 (51.1)	9 (20.9)	0.004	NA	NA

APACHE II—Acute physiology and chronic health evaluation II; ICU—intensive care unit; IQR—interquartile range; NA—not applicable; SD—standard deviation; SOFA—Sequential organ failure assessment. Statistical tests performed for comparison of characteristics of critically ill patients with Candida spp. respiratory isolation, treated or not treated with antifungals, included Student’s t-test for comparison of age, Mann-Whitney for comparison of APACHE II and SOFA score, duration of ventilation and duration of stay in ICU and the hospital, while Fischer’s exact test was used for comparison of proportions of male, prior hospitalization, prior antimicrobial use, tracheostomy, positive blood cultures, ICU and total hospital mortality.

shows the characteristics of patients with a respiratory sample culture positive for Candida spp. Positive cultures derived from bronchial secretions in 53 patients (58.9%), BAL in 31 (34.4%) and sputum culture in 6 (6.7%).

Table 2. Microbiology of isolated fungi from respiratory specimens.

Species	n (%)
C. albicans	60 (66.7)
C. tropicalis	14 (15.6)
C. parapsilosis	6 (6.7)
C. glabrata	4 (4.4)
C. lipolytica	2 (2.2)
C. krusei	2 (2.2)
C. lusitaniae	1 (1.1)
C. dubliniensis	1 (1.1)

shows the isolated fungal species and

Table 3. Microbiology of isolated bacteria from respiratory specimens.

Species	n (%)
Acinetobacter baumannii	40 (44.4)
Pseudomonas aeruginosa	30 (33.3)
Klebsiella pneumoniae	12 (13.3)
Staphylococcus aureus	6 (6.7)
Proteus mirabilis	5 (5.6)
Enterococcus faecalis	3 (3.3)
Morganella morganii	3 (3.3)
Corynebacterium spp.	2 (2.2)
Enterobacter aerogenes	2 (2.2)
Enterobacter cloacae	2 (2.2)
Escherichia coli	2 (2.2)
Haemophilus influenzae	1 (1.1)
Moraxella catarrhalis	1 (1.1)
Staphylococcus saprophyticus	1 (1.1)
Stenotrophomonas maltophilia	1 (1.1)
Streptococcus pneumoniae	1 (1.1)
Proteus vulgaris	1 (1.1)
Staphylococcus haemolyticus	1 (1.1)

shows the isolated bacterial species from the respiratory specimens.

Among all patients, 47 (52.2%) were treated with antifungals during their hospitalization around the time of Candida spp. isolation. Antifungals used were fluconazole in 15 (31.9%), micafungin in 15 (31.9%), anidulafungin in 13 (27.7%), caspofungin in 9 (19.1%), amphotericin B in 4 (8.5%) and voriconazole in 1 (2.1%). As shown in Table 1, where a comparison of characteristics of patients treated with antifungals with those not treated with such agents is presented, age, gender, previous antimicrobial use and hospitalization, tracheostomy, bacteremia, duration of stay in the hospital and ICU mortality were similar in the two groups. Patients treated with antifungals had higher acuity of illness than those not treated, as they had higher APACHE II and SOFA score, longer duration of stay in the ICU, more days on ventilator and higher total mortality during the hospitalization. As shown in

Table 4. Logistic regression analysis of overall mortality during hospitalization.

	Univariate analysis p	Multivariate analysis p	OR (95% CI)
Female gender	0.030	0.296	1.881 (0.575-6.159)
Age (per year)	0.060	0.223	1.026 (0.984-1.069)
BMI (per kg/m2)	0.010	0.114	1.071 (0.984-1.167)
APACHE II (per unit)	0.121	0.589	0.977 (0.900-1.062)
SOFA (per unit)	0.175	0.430	1.074 (0.899-1.283)
Antifungal use	0.003	0.010	4.104 (1.403-12.008)

APACHE II—Acute physiology and chronic health evaluation II; BMI—body mass index; SOFA—Sequential organ failure assessment.

, a multivariate logistic regression analysis identified antifungal use to be independently associated with total mortality during hospitalization.

Discussion

The results of this study have shown that about half of the patients with an ICU-acquired RTI and Candida spp. isolation from their respiratory secretions were treated with antifungals. These patients had a higher acuity of illness than those not treated with such agents. They had higher SOFA and APACHE II scores on admission, longer duration of ventilation, longer stay in the ICU and higher total mortality during hospitalization. These data complicate the present analysis, as patients in a more critical condition could be treated with antifungals more often. Furthermore, treatment with antifungals and longer duration of ventilation were independently associated with increased total mortality during hospitalization.

The importance of Candida spp. isolation from the respiratory tract of critically ill patients with ICU-acquired RTI is unclear. It has been shown that isolation of Candida spp. is associated with higher mortality, increased duration of ventilation and increased ICU and hospital stay [3,9,21]. The role of bacterial-fungal interactions in the lung microbiome has been shown to be important. There are studies suggesting higher probability of acquiring VAP in ventilated patients, if Candida spp. has been isolated from their respiratory tracts [9,22]. To that end, colonization by C. albicans may be associated with impairment of macrophage function and immune dysregulation leading to increased likelihood of development of bacterial pneumonia as shown in rat models [6,7]. Moreover, quorum sensing between Candida spp. and bacteria within the lung microbiome can lead to increased alveolar invasiveness by bacteria, such as A. baumannii and increased likelihood of infection [5]. Furthermore, existence of biofilms may lead to development of antimicrobial resistance of bacteria. For example, S. aureus can form microcolonies on the fungal biofilm of C. albicans, when grown together, leading to polymicrobial biofilm with different phenotypes and single cell membranes, suggesting that S. aureus could be encapsulated in the matrix produced by C. albicans. The result is an increased resistance of S. aureus to vancomycin [23,24]. However, whether antifungal treatment can reduce mortality, duration of ventilation, or ICU and hospital stay is disputable [15,16,21].

It is common practice to treat critically ill patients with Candida spp. isolation from their respiratory tracts with antifungals [21]. In the present study patients treated with antifungals differed from the ones not receiving such agents, since those receiving antifungals had higher scores of clinical severity. Thus, the finding of increased total hospital mortality in the population treated with antifungals could be, in theory, a misinterpretation, since patients being in worse clinical condition could had received these agents more often. To that end, a multivariate regression analysis was conducted identifying antifungal use as an independent factor associated with total hospital mortality.

The unexpected finding of association of antifungal use and higher mortality in critically ill patients with Candida spp. isolated from their respiratory samples is in contrast to the results of other studies identifying a protective role of antifungals in critically ill patients colonized with Candida spp. [25]. For example, Zhang et al., in a retrospective study, found that treatment of respiratory Candida spp. colonization in critically ill mechanically ventilated patients was associated with a reduced multidrug-resistant (MDR) organism isolation ratio, duration of mechanical ventilation, incidence of VAP, length of ICU stay and total length of hospital stay. However, that study included only mechanically ventilated patients, who did not necessarily have an ICU-acquired RTI, as in the present study. In another study, by Lindau et al., antifungal treatment in critically ill patients with Candida colonization did not affect mortality, while SOFA score was an independent factor associated with mortality [26].

The present study has some limitations, such as the relatively small sample size, and the fact that it was performed in a single center, and this may limit generalization of its results. Furthermore, administration of antifungals was based on clinicians’ discretion, which may have led patients in worse clinical condition to be treated with such agents more often. Moreover, the use of antifungals was reported for any reason ±10 days around Candida spp. isolation, which may imply that these agents may have been used, in a very small number of cases, for other reasons. Finally, the observational retrospective design of the study may limit the ability for a causal analysis of the effect of antifungal therapy on clinical outcome.

Conclusions

In conclusion, herein, it was found that critically ill patients with ICU-acquired RTI having Candida spp. in their respiratory specimens, treated with antifungals, did not have any clinical benefit. Moreover, they may even have had higher total hospital mortality, even though sicker patients received antifungals more often. Clinicians caring for such patients should be hesitant to treat them with antifungals, especially when an absolute indication is not present.