The soluble mannose receptor (sMR/sCD206) in critically ill patients with invasive fungal infections, bacterial infections or non-infectious inflammation: a secondary analysis of the EPaNIC RCT

Critical care medicine has evolved dramatically over the last decades, reducing mortality of patients with life-threatening diseases. Nevertheless, long-term outcomes remain poor and patients surviving the acute phase of critical illness are at risk for several complications, among which severe nosocomial infections [1]. In particular, invasive fungal infections (IFI) have emerged in this population and have been associated with an increased morbidity and mortality [1, 2]. Timely initiation of appropriate antifungal treatment is considered an important prognostic factor [3]. However, classical diagnostic tools such as culture and biopsy are invasive and time-consuming and show low sensitivity [4]. Biomarkers, such as serum fungal cell wall components, can be detected rapidly but have a variable sensitivity and a low specificity, especially in intensive care unit (ICU) patients [4, 5]. Moreover, empirical treatment in high-risk patients does not seem to improve clinical outcome [6, 7]. Hence, there is an increasing need for novel and accurate diagnostic tools for IFI during critical illness.

The innate immune system recognises fungal species by pathogen recognition receptors, such as dectin-1 and the mannose receptor (MR or CD206) [8]. The MR is an endocytic receptor present in macrophages, dendritic cells, and endothelial cells and is transported to the cell surface during inflammation. It contains several extracellular domains, each having their own binding capacities to sulfated glycans, carbohydrates, collagens, allergens, and pathogens [9]. Binding of a ligand to the receptor probably stimulates metalloproteinases to cleave the cell-bound receptor into a soluble form, which is shed into the circulation as soluble MR (sMR) [8]. This process is, among others, driven by dectin-1, which is activated by exposure to beta-d-glucan from the fungal cell wall [10]. The sMR shedding is documented in vitro in macrophages exposed to several fungal pathogens such as Candida albicans, Aspergillus fumigatus and Pneumocystis carinii [10, 11]. Elevated serum sMR has also been documented in patients with liver disease, sepsis, invasive pneumococcal disease and malignancies [12‐18].

As MR expression and shedding increase when macrophages are exposed to fungi, we hypothesised that serum sMR concentration could be a valuable tool to diagnose IFI early in critical illness.

The MR is a marker of macrophage activation that is released from the human cell surface and shed into the circulation as sMR with exposure to several stimuli, among which fungal cell walls components [9‐11]. In this study, we investigated whether serum sMR concentration is a clinically valuable diagnostic tool for IFI in ICU patients. We found that sMR concentrations are higher in patients with IFI as compared with patients suffering from a bacterial infection or non-infectious inflammation. The higher concentrations were seen as from 5 days before the IFI was diagnosed by the treating physician. This early rise in sMR concentration may be explained by an association between macrophage activation and the risk to develop IFI and/or the presence of a subclinical infection before clinical suspicion arises. The latter appears likely as classical methods used to confirm diagnosis and initiate treatment are time-demanding and are known to delay treatment initiation [4].

However, also severely critically ill patients who did not suffer from IFI showed elevated serum sMR concentrations, which suggest that sMR shedding from macrophages is rather non-specific. Consequently, the discriminative power appeared to be weak due to the lack of specificity given that patients suffering from a bacterial infection revealed increased serum sMR concentrations. Hence, the use of serum sMR concentration cannot be recommended to diagnose IFI, as its discriminative properties were poor among the studied ICU patients.

Indeed, increased serum sMR concentrations have been documented in patients suffering from a variety of diseases such as liver disease, hyper-inflammation, pneumococcal bacteraemia, malignancies and sepsis [12‐18]. We evaluated the independent association between these characteristics and sMR concentration with the use of multiple linear regression analysis.

Severity of illness, as reflected by the APACHE-II score, was independently associated with sMR concentration, with increasing concentrations for higher disease severity. This is expected, as higher sMR concentrations have been shown for ICU patients as compared with healthy controls and with patients admitted to a medical ward [18]. Indeed, severe illness causes inflammation and tissue damage, activates macrophages and as such induces non-specific shedding of the receptor. Sepsis upon admission was also significantly associated with higher serum sMR concentration, confirming findings from previous studies [13‐15].

Remarkably, withholding supplemental PN until beyond 1 week in ICU appeared to have a significant impact on serum sMR, with lower concentrations when PN was postponed. This may suggest a reduced macrophage activation or a reduced fungal/bacterial load when omitting PN. Previous research has shown an association between the sMR levels and mortality, with higher mortality when sMR increased [14, 15]. The lower sMR concentrations could be an important factor in the improved outcome that was observed in the Late-PN patients in the EPaNIC randomised controlled trial [19]. As far as we know, there are no data on the impact of PN on macrophage activity in humans, so further research is necessary to investigate this possibility.

Another factor significantly associated with a lower serum sMR concentration was emergency admission. However, this finding is difficult to interpret given that more than 95% of the studied patients were admitted non-electively.

In invasive pneumococcal disease, serum sMR concentrations have been shown to be lower in patients older than 75 years as compared to the younger patients [14]. Our studied population was younger, with a median age of 65 years, and we did not find an association between age and serum sMR concentration. Also, history of malignancy in the studied ICU patient cohort was not associated with sMR concentration, unlike in previous studies that reported elevated serum sMR in patients with gastric cancer and multiple myeloma [16, 17]. However, serum sMR concentrations documented in these patients were much lower than serum sMR concentrations that we documented in the cohort of ICU patients suggesting that critical illness is a stronger inducer of MR shedding than cancer.

Our study has some limitations. First, the groups were selected by propensity score matching to avoid confounding. Indeed, patients acquiring an IFI are severely ill and must be compared to equally sick patients. Although groups were well matched for several a priori defined risk factors, other unknown characteristics may still have differed. Second, patients who did not acquire an infection during ICU stay were matched to the infected patients for a similar ICU day. This was most appropriate because duration of illness may otherwise be a confounder. However, we cannot exclude that serum sMR may have declined in patients with non-infectious inflammation due to recovery. Third, the group of patients with IFI consisted of patients with proven and probable IFI. When comparing sMR concentrations in patients with proven invasive aspergillosis, probable invasive aspergillosis, candidaemia and other invasive candida diseases, sMR appeared numerically higher in patients with proven invasive aspergillosis and with candidaemia as compared to other patient groups. However, no statistical analyses could be performed given the too small numbers for this comparison. The lower serum sMR concentrations in patients with probable as compared with proven invasive aspergillosis and in patients with invasive candidiasis as compared with candidaemia could have been a consequence of less severe infections. However, misdiagnosis could also play a role given that available diagnostic tools are lacking sensitivity and specificity [4]. Theoretically, patients who did not survive ICU stay may have had an unrecognised IFI. Routine post-mortem examinations were not available to exclude this possibility. Fourth, this is a secondary analysis of the EPaNIC RCT that included many patients from a surgical ICU. However, the sickest patients were selected by propensity score matching and the final cohort consisted of a mixed ICU population with 32.2% of all patients admitted to a medical ICU.

Our study is the first clinical study to investigate serum sMR concentrations in patients with IFI. Although cell culture experiments have shown an increased shedding of MR in different cell types after exposure to fungi, studies investigating this aspect in patients were lacking. We found that the serum sMR concentration increases significantly in critically ill patients with IFI as compared with those without IFI, and this already 5 days before clinical diagnosis. We showed a significant impact of sepsis upon admission and of severity of illness on serum sMR, and a significant lowering of serum sMR with postponing administration of PN.

Serum sMR concentrations are higher in ICU patients who suffer from an IFI than in those with a bacterial infection or with non-infectious inflammation. This clinical finding confirms earlier experimental evidence for a rise in MR shedding in response to fungal exposure. However, the power of serum sMR concentrations to discriminate IFI from other infections or other causes of inflammation was insufficient to advise it as a clinically reliable diagnostic tool.