In non-immunocompromised intensive care unit (ICU) patients Candida isolation from various sites is common [1]. The association with the risk of invasive candidiasis (IC) is the rational for the use of predictive scores which include the colonization status.

However the meaning of Candida spp. recovery from the lung is definitely more intriguing. Apart from IC episodes occurring in deeply immunosuppressed subjects who may develop true fungal pneumonia, Candida respiratory tract (RT) isolation should not be considered a marker of lung infection. This issue was addressed by Meersseman et al. [2] who did not identify any autopsy-proven case of Candida pneumonia among 135 autopsies with histopathological evidence of bacterial infection and high rate of Candida isolation from endotracheal aspirate (EA) and/or bronchoalveolar lavage (BAL) (57 %).

However there is growing convincing evidence, based on animal studies and human observations, that Candida spp. is not definitely an innocent bystander in the respiratory tract of ICU ventilated patients [3–13] (Table 1). Beta-glucan (BG), a component of yeast cell wall, may act as a lung proinflammatory agent causing alveolar macrophage and neutrophil dysfunction. Additionally, within the environmental biofilm there is a strong interplay, through quorum-sensing (QS) molecules, between Candida and both Gram-positive and Gram-negative bacteria [14]. Live Candida albicans instillation in rats has been observed to increase the susceptibility to develop experimental Pseudomonas aeruginosa (PA), Escherichia coli (Ec), and Staphylococcus aureus (Sa) pneumonia, fostering the production of lung inflammatory cytokines (tumor necrosis factor alpha [TNF-alpha], interleukin-6 [IL-6], and interferon-gamma [INF-γ]) and inhibiting alveolar macrophage phagocytosis [3, 5]. From a clinical viewpoint, Candida airway colonization has been shown to be associated with prolonged duration of mechanical ventilation, ICU/hospital length of stay, and increased mortality [8–11]. One of the first reports of this possible relationship dates back to almost 10 years ago, when Azoulay et al. [6] identified Candida bronchial isolation as an independent risk factor for the development of PA ventilator-associated pneumonia (VAP) (9 vs 4.8 % in non-colonized patients, p = 0.048). These data were strengthened just 1 year later by Nseir et al. [7] who observed a protective role of antifungal treatment on PA VAP occurrence in a cohort of patients with Candida tracheobronchial colonization. However, recently, the use of nebulized amphotericin B (NAB) was not able to provide any clinical improvement, albeit increasing the rate of Candida decolonization (adjusted HR 2.2; 95 % CI 1.6–3) [13]. Hence the clinical usefulness of respiratory tract Candida colonization (CC) eradication in mechanically ventilated patients is still a matter of debate and the paper by Albert et al. [15] in the current issue of Intensive Care Medicine is the first interventional trial aiming to describe the biochemical and clinical effect of antifungal treatment in ICU patients with VAP and RT Cc. In this double-blind, placebo-controlled, multicenter study 60 patients were enrolled: 29 in the placebo group and 31 in the antifungal strategy group. A comparative observational cohort of 29 patients with VAP but without Candida respiratory colonization was also included. No differences between the two arms were found in any of the investigated clinical outcomes: maximum and delta Sequential Organ Failure Assessment (SOFA) score, MV- and ICU-free days, subsequent ICU-acquired infections, and day-28, day-90, and hospital mortality. In addition, the inflammatory profile (TNF-alpha, IL-6, IL-8, IL-10, IL-1B, C-reactive protein [CRP], procalcitonin [PCT], intestinal fatty acid binding protein (iFABP), and lipopolysaccharide [LPS]-stimulated TNF-alpha) was similar between the antifungal and the placebo group, not being influenced by the assigned treatment. On the other hand patients harboring Candida spp. in the RT, compared with the observational group, showed higher TNF-alpha values (21.8 ± 23.1 vs. 12.4 ± 9.3 pg/mL, p = 0.02) and lower LPS-induced TNF-alpha production capacity (854.8 ± 855.2 vs. 1,559.4 ± 1,290.6 pg/mL, p = 0.01). Unfortunately, because of the slow enrollment rate (0.6 patients per month per site for the intervention groups), the trial was interrupted long before the completion of the planned sample size (120 patients for the randomized trial and 40 for the observational cohort). The authors’ conclusion excluded the feasibility of a larger phase 3 trial aiming to assess the potential benefits of the antifungal treatment in patients with concomitant RT Candida spp. and VAP but some aspects of this interesting study deserve further investigation.

Table 1 Main animal and human studies on RT Cc and bacterial pneumonia relationship
Full size table

First, one important issue might be the choice as inclusion criteria of clinically suspected VAP (csVAP), rather than microbiologically confirmed cases. Only in 14 out of 60 randomized patients (six in the placebo arm and eight in the intervention arm) were the respiratory specimens positive for pathogenic bacteria and only two Pa isolates were observed.

Clinical diagnosis of VAP is challenging: in ventilated patients chest X-ray images may be frequently misleading and about 30 % of ICU-acquired pneumonia remains without etiology. The authors previously described that patients with RT Candida colonization and csVAP may experience similar inflammation burden and worse outcome than patients with bacterial infections [11]. Airway Candida colonization is associated with pulmonary inflammation and consequent cellular immune dysfunction, but the role of biofilm and fungal-bacterial cross-talk should not be neglected.

Candida biofilm consists of a network of yeasts, hyphae, and/or pseudohyphae embedded in a matrix of polysaccharides, proteins, and other undefined components. This status results in the protection of the fungi from the host defenses and the development of antifungal drug resistance. Many bacteria are able to produce biofilm and studies using electron microscopy showed that Pa, Enterobacteriaceae (including Ec), Sa, and Candida spp. are the most common pathogens collected from artificial airways of mechanically ventilated patients (Fig. 1) [16]. Despite billions of years of cross-kingdom interaction, little is know about Candida–bacteria interactions within the biofilm. A wide range of metabolic processes and cell-to-cell communications through QS molecules are the basis of both synergistic and antagonistic interactions. Interestingly, in a prospective observational study, RT Cc was identified as an independent predictor of multidrug-resistant bacteria airway isolation in patients with suspected VAP [12]. As a result of the exiguous number of microbiologically confirmed VAP in the CANTREAT study, no conclusions may be drawn about these phenomena.

Fig. 1
figure 1

Scanning electron microscopy of an uncoated endotracheal tube after extubation (a luminal surface, b bacterial biofilm) [16]

Full size image

Second, Albert et al. [15] did not provide details on biofilm production of isolated Candida spp. A recent survey of ICU patients with IC reported in 297 Candida isolates, percentages of biofilm formation ability ranging between of 20 and 50 % and higher mortality rates in patients with candidemia may be observed in the presence of high producers strains [17]. In the CANTREAT trial, patients in the intervention group received anidulafungin for a mean of 5.9 ± 3 days but only 22.6 % were not sequentially switched to fluconazole. Unlike fluconazole, echinocandin and liposomial amphotericin B are highly active against both planktonic and biofilm Candida spp. but, despite the high 72 h Candida eradication rate, we are not aware of the potential beneficial anti-biofilm effects derived from the use of anidulafungin.

Third, the authors investigated in detail the inflammatory profile of enrolled subjects. Patients with csVAP and RT Cc were distinctly characterized by an increased inflammatory status and a relative immunosuppression; however, this pathophysiological condition did not change over antifungal treatment, as well as BG serum levels. It is noteworthy that the average BG levels were fairly low and mainly in the range of possible false positive results [18]. In addition, the fungicidal properties of anidulafungin might have determined an early BG release from the yeast cell wall, influencing the kinetics of this biomarker and of the local/systemic inflammatory response.

Finally, the very restrictive inclusion criteria together with a relevant percentage of excluded patients (72 out 133, 54.1 %) contributed to the premature interruption of the trial, which did not reach the predetermined sample size.

A definitive conclusion on the clinical relevance of respiratory tract Candida colonization has not yet been reached, and at present data are still insufficient to recommend the routine use of antifungals as decolonizing agents in non-immunocompromised patients.