Invasive fungal infections in patients with liver disease: immunological and clinical considerations for the intensive care unit

Abstract

Patients with liver disease in the intensive care unit (ICU) face a unique susceptibility to infection due to the complex immune dysfunction resulting from hepatic failure. Bacterial infections are commonly present in these patients upon arrival to the hospital, often being the primary reason for ICU admission. In contrast, invasive fungal infections (IFIs) afflict a smaller percentage of patients and are usually discovered in the course of the ICU stay. IFI diagnosis in the ICU, particularly in patients with liver disease, is often delayed or overlooked, contributing to the extremely high ICU mortality associated with IFI in these patients despite the availability of effective (and largely safe) antifungal therapy. Thus, to improve outcomes, it is crucial for intensive care clinicians to be vigilant for IFIs in patients with liver disease. This review aims to contribute to the intensive care literature in this regard. We begin with an overview of normal antifungal immunity followed by a summary of how it may become compromised in the setting of hepatic dysfunction. Next, a general discussion of IFIs in liver disease is presented and then the three most relevant fungal pathogens, namely Candida, Aspergillus, and Cryptococcus, are individually examined. This review concludes by highlighting key knowledge and practice gaps that require attention by the scientific and clinical communities in the coming years.

Take-home message

Patients with liver disease constitute a growing intensive care unit population that poses many unique challenges, one of which is their susceptibility to infection, in particular fungal infection. Liver disease compromises not only local hepatic immune function but also has an impact on systemic antifungal defenses, so it is essential for intensive care providers to recognize this vulnerability and maintain a high index of suspicion for invasive fungal infection in critically ill patients with hepatic failure. The contemporary approach to the diagnosis and treatment of invasive fungal infection does not require major modifications in the presence of liver disease.

Introduction

Patients with liver cirrhosis have constituted a progressively higher proportion of intensive care unit (ICU) admissions in recent decades. For example, the percentage of such patients nearly doubled from 1.6% to 3.1% in the UK between the years 1998 and 2012 [1]. In the contemporary experience of a major US tertiary referral center, the hospital course of 20% of patients with cirrhosis admitted through its Emergency Department included an ICU stay—most commonly for septic shock [2]. In the Western world, alcohol use and nonalcoholic fatty liver disease are becoming increasingly prevalent etiologies of cirrhosis while cases caused by viral hepatitis are declining [3].

Because liver disease leads to both local and systemic immunodeficiency, infection is a prominent reason for ICU admission in this patient population and is a frequent precipitant of acute-on-chronic liver failure (ACLF) [4]. Most infections in patients with liver disease are bacterial, with invasive fungal infections (IFIs) constituting a significant minority. In a multi-center US cohort, for example, IFIs were prospectively identified in 134 out of 2743 inpatients with cirrhosis for an overall incidence of 3% and a 13% proportion of all infections [5]. IFIs in this patient population are important to the intensive care community because they occur in the ICU (more commonly) or lead to ICU admission (less commonly) and are strikingly associated with short-term mortality. Over the years, several studies have shed light on the subject of IFIs in patients with liver disease, but rarely has this topic been reviewed with the intensive care clinician in mind. Even less often has this topic been approached from the perspective of antifungal immunity in terms of both its normal mechanisms and their impairment in the setting of hepatic dysfunction. Many important insights about antifungal immune defects in liver disease have been added to the scientific literature in recent years.

The aim of the present review is to provide the reader with a clinically oriented synopsis of antifungal immunity in health and in liver disease followed by a discussion of the three fungal pathogens most relevant to the intensive care of the liver disease patient: Candida, Aspergillus, and Cryptococcus. Although Mucorales and the endemic fungi (e.g., Histoplasma, Blastomyces, Coccidioides) can be encountered in critically ill patients with liver disease under certain circumstances [6–9], a link between these infections and hepatic dysfunction has not been definitively established.

Basic principles of normal antifungal immunity

Innate immunity

After mucosal and endothelial barriers, the human host’s innate immune system serves as the next line of defense against IFIs (Fig. 1). The innate immune system includes phagocytic cells, chief among them being neutrophils and macrophages. Their role is to recognize fungal pathogens as foreign and neutralize them by means of phagocytosis and intracellular or extracellular killing, which are complex biological functions that can be activated by a variety of signaling cascades triggered by microbe recognition [10]. Unlike cells of the adaptive immune system, innate immune cells are available for immediate response. The structural components of fungi that enable phagocytes to identify them as such are collectively termed pathogen-associated molecular patterns (PAMPs). Often, fungal cell wall components, such as β-glucans found in Candida and Aspergillus, galactomannan found in Aspergillus, and mannans found in Candida, serve as PAMPs. Encapsulated fungal pathogens, Cryptococcus being one example, are adept at evading recognition because their capsule conceals their PAMPs from immune surveillance [11, 12].

Fig. 1.

Diagram of the normal innate and cellular adaptive immune responses to fungal organisms highlighting the central role of macrophages and dendritic cells (DC). Should fungi succeed in breaching the host’s external barriers, the first immune cells they encounter as part of the innate response are the professional phagocytes residing in tissues: macrophages and DCs. Macrophages are responsible for both phagocytosis and, through signaling based on molecular recognition as detailed in the figure, for elaborating pro-inflammatory cytokines, such as TNF-α and IL-1β, which attract blood-borne effector cells (neutrophils and monocytes) to the site of invasion. Likewise, DCs contribute to the innate immune response not only through phagocytosis but also by stimulating natural killer cells, which, once activated, release cytotoxins for direct pathogen elimination. The other critical function of DCs is their migration to surrounding lymph nodes, where they promote the differentiation of naïve CD4 + lymphocytes toward, among others, Th1 and Th17 phenotypes. The former pathway leads to enhancement of the fungicidal function of macrophages, which is critical for defense against Cryptococcus in particular because of this pathogen’s ability to evade recognition and destruction by the innate immune response thanks to the shield provided by its polysaccharide capsule. The latter pathway, mediated by IL-17A and IL-17F, provides protection at mucosal surfaces by stimulating epithelial cells to produce antifungal peptides and through recruitment of neutrophils. DC dendritic cell, MHC major histocompatibility complex, MyD88 myeloid differentiation response 88, NK natural killer, PAMP pathogen-associated molecular pattern, PRR pattern recognition receptor, SYK spleen tyrosine kinase, TCR T cell receptor, TLR toll-like receptor

Phagocytic cells are equipped with various pattern recognition receptors (PRRs) capable of binding to PAMPs (reviewed in detail elsewhere [13]); for antifungal host defense, a few specific classes of PRRs deserve particular mention. The most important family of PRRs in this context is that of C-type lectin receptors (CLRs), of which Dectin-1 is the best studied. Dectin-1 recognizes β-glucan, a ubiquitous fungal cell wall component found in many genera. Upon engagement of Dectin-1 by β-glucan, a tyrosine kinase known as Syk triggers an intracellular signaling cascade at the center of which is the caspase recruitment domain-containing protein-9 (CARD9). This pathway culminates in the activation of nuclear factor-κβ (NF-κβ), production of pro-inflammatory mediators, and recruitment of neutrophils and monocytes to the site of infection [14, 15]. The critical role of CARD9 in human antifungal immunity is illustrated by studies demonstrating a predisposition to IFIs of human hosts with inherited CARD9 deficiency, the only known inherited immunodeficiency to date that causes susceptibility specifically to fungal infections [16–18].

Toll-like receptors (TLR) 2 and 4, widely expressed on macrophages, play a vital role guarding against invasion by Candida owing to their ability to detect both structural (e.g., mannoproteins) and secretory (e.g., Sel1) proteins characteristic of this genus [19]. The mannose receptor can likewise sense Candida by the mannan component of the glycoproteins that make up the outer layer of its cell wall. TLR engagement leads to signaling via an adaptor protein called myeloid differentiation primary response 88 (MyD88) that activates NF-κβ, which primes the production of downstream pro-inflammatory mediators among which are interleukin-1β and tumor necrosis factor-α (TNF-α). The involvement of these proteins in antifungal defense helps explain why patients receiving therapy with medications that inhibit NF-κβ activation (i.e., glucocorticoids) and those that target TNF-α are at increased risk for IFIs. It is worth mentioning that the Fcγ receptor on the surface of phagocytes best known for binding opsonized pathogens as part of the adaptive immune response also serves as a co-receptor in the innate antifungal immune response owing to its ability to bind mannan and its derivatives [20].

A component of the innate immune response that is also crucial for adaptive immunity to fungal infection is the dendritic cell (DC). DCs are among the professional phagocytes capable of neutralizing fungi on initial contact, but they ensure preservation of fungal peptides for their primary role as antigen presenting cells in lymphoid tissues [21]. Importantly, DCs potentiate the cytotoxic activity of natural killer (NK) cells [22]. Considered a type of innate lymphoid cell (ILC), this small lymphocyte subset (15% of the total population) has been increasingly appreciated as central to frontline fungal surveillance and elimination [23]. In general, ILCs are local responders to mucosal damage found in abundance at sites such as the lung and intestine, the two most important portals of entry for IFIs. In murine models, ILCs that secrete IL-17 have been demonstrated to play a role in fungal control not only in mucosal [24] but also in systemic [25] candidiasis. Finally, the complement system—previously thought to be strictly an antibacterial arm of innate immunity—has recently been shown to contribute to protection against systemic Candida infection [26].

Adaptive immunity

The adaptive immune system consists of two parts: humoral and cellular. The central cell of the former is the B lymphocyte responsible for producing antibodies, which coat extracellular pathogens and thereby facilitate their clearance. The latter is mediated by CD4 + T lymphocytes (i.e., helper T cells) following antigen presentation by phagocytes, such as macrophages and DCs. The helper T cell population consists of multiple subtypes that vary according to their cytokine profile and thus the eventual effector cells they stimulate. Cell-mediated immunity is the more important adaptive arm for antifungal defenses. Specifically, the Th1 helper T cell response is characterized by the elaboration of, among other cytokines, interferon-γ (IFN-γ) and TNF-α. This leads to activation of macrophages and CD8 + T lymphocytes (i.e., cytotoxic T cells), enabling them to eliminate infected cells more effectively [27]. The Th1 pathway is also viewed as a form of protection against disseminated infection by fungal pathogens, Cryptococcus notable among them, that are capable of survival and proliferation within the macrophages they encounter as part of the innate immune response [28, 29]. The Th17 pathway, acting via IL-17A and IL-17F, plays a vital role in antifungal defense at the mucosal interface by recruiting neutrophils and by stimulating the production of antimicrobial peptides that directly inhibit Candida [30]. The importance of this pathway is supported by the clinical observation that inherited IL-17 receptor defects result in chronic mucocutaneous fungal syndromes [31–33].

Immunological defects in liver disease

The liver is a highly immunologically active organ (Fig. 2). It is not surprising, therefore, that infection is a prevalent cause of death in patients with liver failure [34]. The impact of liver disease on immune system integrity can be separated into two components: local and systemic. The local liver immune function that is impaired in failure states is microbial filtering of the splanchnic venous drainage delivered by the portal vein. Kupffer cells (KC) are local macrophages found in abundance in the liver and tasked with antimicrobial surveillance of blood flowing through hepatic sinusoids. Using elegant murine experiments, it has been shown that KC depletion leads to reduced clearance and increased dissemination of Cryptococcus neoformans and Candida albicans, indicating that KCs are a protective mechanism against systemic fungal spread [35]. In the setting of C. neoformans challenge, the protective effect of KCs on pathogen control was observed even after intratracheal inoculation, demonstrating the importance of liver integrity for containment of fungal infections acquired via the respiratory route. From a mechanical perspective, advanced liver disease is associated with abnormal intestinal mucosal integrity, promoting translocation of gut-resident pathogens, prominent among them being Candida spp. [36].

Fig. 2.

The liver is a vital cog in human immune defenses against infection, both locally and systemically, as illustrated in this diagram. The main immune function of the liver is antimicrobial surveillance of splanchnic blood returning from the portal circulation and flowing through liver sinusoids into hepatic venules. This flow is depicted from right to left across the central part of the diagram. Kupffer cells (KC) are macrophage derivatives that reside within hepatic sinusoids where they serve as the primary phagocytes and are also very important antigen presenting cells (APCs). KCs as well as dendritic cells (DC), the other prominent APCs residing in the liver, are major producers of both pro- (e.g., TNF-α) and anti-inflammatory cytokines (e.g., IL-10) and are thus instrumental in calibrating the liver’s immune response. Unlike KCs, DCs are not confined to the lumen of hepatic sinusoids, so they are able to traverse the Space of Disse and migrate to locoregional lymph nodes, where they initiate an adaptive immune response by priming naïve CD4 + lymphocytes. In the hepatic sinusoids, KCs and DCs interact with a variety of intrahepatic lymphocytes, most numerous among them being natural killer (NK) cells. Even structural cells in the liver, namely hepatocytes and sinusoidal endothelial cells, play a number of immunological roles, including elaboration of cytokines, release of soluble pattern recognition receptors, and phagocytosis. The major liver-resident cells with immune functions are depicted in the top left section of the diagram. Aside from disrupting local antimicrobial defenses, liver cirrhosis also negatively impacts the systemic immune response as illustrated in the bottom right section of the figure. Despite an overall hyperinflammatory systemic milieu, specific circulating immune cells are rendered dysfunctional in liver failure as part of a phenomenon called immune cell paralysis. In this state, neutrophils become less effective with respect to chemotaxis and phagocytosis. The oxidative burst critical for effective pathogen elimination by monocytes is impaired, and both a quantitative and qualitative defect in T cell lymphocyte immunity develops. Figure used with permission of Mayo Foundation for Medical Education and Research, all rights reserved. DAMP damage associated molecular pattern, DC dendritic cell, KC Kupffer cell, L lumen of hepatic sinusoid, LPS lipopolysaccharide, PAMP pathogen-associated molecular pattern, PRR pattern recognition receptor, TLR toll-like receptor

The impact of liver failure on systemic immunity relates to multifactorial immune cell dysfunction associated with that condition as well as to reduced protein synthesis by the liver, which affects production of complement components and levels of soluble PRRs [37]. Liver disease dampens the innate immune response by causing various abnormalities in phagocytosis. Deficiency in chemotaxis [38], oxidative burst [39], cell survival [40], and so-called “swarming” [41] is characteristic of neutrophils in patients with cirrhosis. Neutrophil dysfunction in cirrhosis appears to exist despite a milieu rich in inflammatory mediators, and the culprit factor is likely bloodborne because normal neutrophils become dysfunctional after incubation with plasma from patients with cirrhosis [41, 42]. The exact identity of this factor or factors remains unknown, but a role for ammonia has been postulated [43]. Evidence also points to weakened NK cell cytotoxic activity in patients with cirrhosis [44]. The best described deficit in the cell-mediated arm of adaptive immunity in liver disease has been a quantitative defect in the CD4 + (helper) T cell lymphocyte subpopulation [45, 46]. Multiple explanations may underlie this finding, including reduced thymopoiesis [47], increased splenic sequestration [48], and a blunted response to proliferative signals [49]. On the humoral adaptive side, both opsonization [50] and subsequent uptake by macrophages via their Fcγ receptor [51] have been shown to be impaired in liver disease. It is worth recalling that the Fcγ receptor also plays a role in the innate antifungal immune response as co-receptor with CLRs, such as Dectin-2, Dectin-3, and Mincle [20]. A different angle on the interplay between synthetic dysfunction of the liver and fungal proliferation involves iron metabolism and the protein hepcidin, which is produced by the liver and is normally upregulated in the setting of infection to deprive pathogens such as Candida of access to iron. Results of recently concluded laboratory work suggest that reduced hepcidin levels in the setting of cirrhosis may promote fungal dissemination [52]. For additional details, please see Fig. 2, which illustrates the intrinsic immunological components of a healthy liver that are compromised by hepatic failure and graphically summarizes its detrimental impact on the function of circulating immune cells.

Overview of invasive fungal infections in liver disease

Impaired antifungal immunity is a major reason why patients with liver disease constitute favorable hosts for the replication of fungal pathogens. Whether immunological susceptibility to IFIs is linked to worsening liver function in a continuous manner [e.g., as a function of rising Model for End-Stage Liver Disease (MELD) score] or through a threshold effect (e.g., increased in Child–Pugh class B and C cirrhosis but not in A) is currently unknown. Limited available data point to the possibility of increased IFI risk with higher MELD score [53] and/or ACLF grade [54] and also suggest that complications, such as hepatorenal syndrome [55] and hepatic encephalopathy [56], may be signals of elevated IFI risk in the setting of cirrhosis. In aggregate, liver disease states that appear to render patients especially vulnerable to IFIs are decompensated cirrhosis, ACLF, and severe alcoholic hepatitis treated with corticosteroids (see below).

Unchecked fungal proliferation is but one factor in determining whether eventual clinical IFI will occur; another is the composition of the so-called “mycobiome,” or fungal ecosystem of an individual, which has been shown to be skewed toward pathogenic fungi in a variety of liver diseases [57]. Both of these predisposing factors are promoted by the long-term antibiotic therapy that many patients with cirrhosis receive as prophylaxis against spontaneous bacterial peritonitis. Selection pressure imposed by antibiotics is further accentuated during periods of critical illness since potent treatment of bacterial infection is often part of the ICU management of these patients. While intravenous antibacterial agents broadly active against intestinal flora expectedly promote fungal proliferation, even the narrow-spectrum antibiotic vancomycin has been shown in murine experiments to weaken fungal defenses by suppressing antifungal cytokine production (e.g., IL-17A) by lymphoid cells in the gut [58]. It comes as little surprise, therefore, that the rate of fungal colonization is higher among critically ill patients with cirrhosis than among their counterparts without cirrhosis [59], and in one study mere colonization was observed to confer greater mortality in the former but not in the latter patient population [60]. Fungal colonization is often a precursor of IFIs, which accounts for the danger of colonization, and the conversion from colonization to infection is greatly facilitated in the setting of liver disease [61]. Particular susceptibility appears to occur during prolonged admissions for ACLF that include an ICU stay [5, 62].

The vast majority of IFIs in patients with liver disease are caused by Candida, Aspergillus, and Cryptococcus, listed in order of frequency, and are most often nosocomial complications (Table 1). The burden of IFIs as a whole in patients with liver disease admitted to the ICU is not well established. Analysis of prospectively collected data from a large UK database demonstrated an incidence of IFIs of 1% within the first three days of ICU stay, which was not different from a comparison group of critically ill patients without cirrhosis [59]. On the other hand, an international ICU single-day point-prevalence estimate taken at a median of > 10 days of ICU stay yielded an IFI rate of 13% in patients with cirrhosis, significantly higher than the rate of 8% in those without cirrhosis [63]. These numbers illustrate that IFIs in critically ill cirrhotics are rarely a factor in the early ICU course but are more likely than in the general ICU population to develop at a later period in their stay. Evidence that the above prevalence figure for patients with cirrhosis may be an underestimate comes from a recent autopsy study showing that among 17 cirrhotic patients with IFIs identified postmortem, in only 11 (65%) was there antemortem suspicion for the presence of an IFI [64]. All 6 of the unsuspected IFIs revealed at autopsy in this study were due to Candida spp., with 2 disseminated infections, 2 lung infections, 1 candidemia, and 1 case of esophageal candidiasis. Patients with severe alcoholic hepatitis treated with corticosteroids elicit particular concern for IFIs. In a small sample of 12 biopsy-proven ICU cases of severe alcoholic hepatitis, 8 patients developed IFIs for a staggering incidence of 67% [65], though a larger study [66] reported very few fungal isolates while meta-analysis of randomized trials of corticosteroid therapy for severe alcoholic hepatitis [67] yielded an overall incidence of < 1% (significantly higher in corticosteroid recipients).

Table 1.

Summary of important characteristics of Candida, Aspergillus, and Cryptococcus in liver disease patients for the intensive care unit provider

Pathogen	Form	Portal of entry/reservoir	Usual diagnostic material	Common syndromes	First-line therapy	Special considerations in liver disease
Candida	Yeast	Gastrointestinal	Blood culture Ascites culture	Candidemia SFP	Echinocandin	Most common IFI in liver disease Extremely high mortality with SFP
Aspergillus	Mold	Respiratory	Serum GM BAL culture BAL GM	IPA	Mold-active triazole (e.g., voriconazole)	Reduced serum GM sensitivity compared to neutropenic hosts Hepatotoxicity of voriconazole Newer triazoles appear to be less hepatotoxic
Cryptococcus	Yeast	Respiratory	Blood culture Serum antigen Ascites culture CSF antigen	Meningitis SFP	AmB plus flucytosine (5-FC)	High rate of disseminated infection Survival very poor in disseminated cases Mild transaminitis with 5-FC

AmB amphotericin B, BAL bronchoalveolar lavage, CSF cerebrospinal fluid, GM galactomannan, IPA invasive pulmonary aspergillosis, SFP spontaneous fungal peritonitis

Across studies, the pooled mortality rate associated with all types of IFIs in cirrhosis is 64%, with ICU admission being a risk factor for death [68]. Robust mortality data specific to ICU patients with liver disease and IFIs are lacking. In the aforementioned UK ICU database study, none of the cirrhotic patients with early IFIs died in the hospital, likely a function of the timing of IFI diagnosis and preponderance of compensated cirrhosis [59].

Specific invasive fungal infections in critically ill patients with liver disease

General clinical considerations

For research purposes and also increasingly in clinical practice, IFIs are defined based on the most current iteration of the consensus criteria promulgated by the European Organization for the Treatment of Cancer and the Mycoses Study Group Education and Research Consortium (EORTC/MSGERC). The full version of these criteria published in 2020 addresses molds such as Aspergillus and yeasts, such as Candida and Cryptococcus, in the general population [69], whereas a supplement published in 2021 is focused specifically on invasive candidiasis and invasive aspergillosis in the ICU [70]. Briefly, proven IFI requires either histopathological demonstration of the organism in tissue or its isolation from a normally sterile site. The diagnosis of probable IFI, on the other hand, rests on the triad of a susceptible host, a compatible clinical syndrome and radiological picture, plus a positive result of at least one supportive mycological test. In the 2020 edition, liver disease was not included among the host factors for any IFI, but the 2021 supplement listed decompensated cirrhosis as a host factor for invasive aspergillosis.

Contemporary clinical practice guidelines aimed at the intensive care provider have covered the most salient questions pertaining to the diagnosis and treatment of both invasive candidiasis and invasive pulmonary aspergillosis (IPA) in the critically ill [71–73]. Since a detailed discussion of this subject matter is beyond the scope of the present article, the reader is referred to these guideline documents and to excellent recent reviews [74, 75] for more information. Of note, the guideline authors did not consider patients with liver disease separately because, fundamentally, the diagnostic and therapeutic approach to these patients is similar to what is recommended for other non-oncological patient types. In the special case of decompensated cirrhosis complicated by ascites, peritoneal fluid is available as a unique source of culture material for potential identification of fungi capable of causing spontaneous peritonitis, namely Candida and Cryptococcus. With respect to IPA diagnostics, it is important to be mindful of diminished sensitivity of the serum galactomannan assay in non-neutropenic hosts such as those with liver cirrhosis [76]. The sensitivity of bronchoalveolar lavage fluid testing for galactomannan in critically ill cirrhotics, on the other hand, has been reported to be approximately 90%, which is not inferior to its performance in other populations [53]. In terms of IPA treatment, older triazole antifungals such as voriconazole are associated with clinically significant hepatotoxicity, posing a possible hazard in the setting of cirrhosis (see below and Table 1). Echinocandins, the drugs of choice for invasive candidiasis in the ICU, are not known to be hepatotoxic and thus no special considerations are warranted in liver disease. The same holds true for Amphotericin B (AmB), the cornerstone of therapy for most manifestations of cryptococcosis. Flucytosine (5-FC), the preferred adjunct to AmB for synergy, can induce a transient transaminitis, but clinically apparent liver injury is exceedingly rare; therefore, the omission of 5-FC in the setting of underlying liver disease is not indicated [77].

Invasive candidiasis and candidemia

Candida is a genus consisting of dimorphic fungi normally confined within human mucosal barriers but capable of causing invasive candidiasis in susceptible hosts, most commonly manifested clinically as candidemia. Historically, neutropenia has been the predisposing factor most tightly linked to invasive candidiasis, and cancer patients with chemotherapy-induced neutropenia remain an especially high-risk category. Over time, numerous invasive candidiasis risk factors not involving quantitative immune cell deficiencies or therapeutic immunosuppression have emerged. The list of the most convincingly demonstrated risk factors includes receipt of antibacterial agents, receipt of total parenteral nutrition, abdominal surgery and pathology (e.g., necrotizing pancreatitis), central venous catheterization, and kidney failure, especially if requiring renal replacement therapy [78, 79]. These patient characteristics are very common among the critically ill, often in combination, so it is no surprise that invasive candidiasis is particularly prevalent in ICUs [80]. Both ICU-specific [81, 82] and general [79] clinical prediction rules for invasive candidiasis have been derived using identified risk factors, but none of them incorporates liver disease.

Evidence suggests that the primary reservoir of Candida in the human body is the gastrointestinal tract and spread occurs from there to other common sites, such as skin and the genitourinary system [83]. The hyperpermeable intestinal epithelium of liver disease predisposes to escape of intrinsic Candida spp. into ascitic fluid leading to spontaneous fungal peritonitis (SFP), a Candida infection unique to patients with cirrhosis [84]. Another escape route available to intestinal Candida is the mucosal venous drainage collected by the portal vein. As reviewed above, microbial filtration of portal venous blood is defective in liver disease, facilitating translocation of Candida into the systemic circulation. Impairment of both innate and adaptive immune clearance mechanisms associated with hepatic dysfunction could then be expected to promote fungemia and dissemination of Candida to end organs that are thereby damaged.

Candida spp. account for 7–10% of bloodstream infections among patients with liver cirrhosis, with most cases occurring in the ICU [85–88]. Conversely, approximately 14–18% of candidemic patients in the ICU have underlying liver disease [78, 89–92]. Studies of ICU candidemia have not identified cirrhosis as an independent risk factor for the development of candidemia [69, 71]; liver disease has, however, been linked to higher mortality from ICU candidemia [91, 93]. A contemporary meta-analysis has estimated the crude mortality of critically ill cirrhotic patients with invasive candidiasis (candidemia and SFP) to approach 80%, with SFP being associated with worse survival than candidemia [94]. The development of septic shock in the setting of invasive candidiasis portends a particularly poor outcome in patients with cirrhosis [88] as it does in the general ICU population [91]. Based on limited data, cirrhosis does not appear to be an independent risk factor for ICU candidemia due to non-albicans Candida species [90].

Invasive pulmonary aspergillosis

Aspergillus is an environmental mold capable of causing human infection in the setting of compromised innate immunity, particularly in patients with numeric or qualitative defects in neutrophils. It is not unusual for the lower respiratory tract to be colonized by Aspergillus, but progression to invasion requires a local or systemic immunological defect. Since the mode of acquisition is inhalational, the most common invasive disease caused by Aspergillus is IPA, making it a prevalent problem in critical care practice. The classical risk factors for IPA are chemotherapy-induced neutropenia in patients with hematological malignancy and post-transplant immunosuppression, particularly when it is used to treat graft-versus-host disease complicating allogeneic stem cell transplantation. The list of predisposing host factors for IPA has been expanding over time. Whereas twenty years ago, the association between liver disease and IPA may have been limited to case series, and contemporary guideline documents have listed it as a formal risk factor [95].

The prevalence of IPA in critically ill patients with cirrhosis is difficult to determine accurately given the many challenges of establishing this diagnosis in the ICU, making IPA likely an underdiagnosed ICU infection. One such challenge that is specifically worth noting in this review is that IPA in critically ill patients with liver disease almost never manifests on chest computed tomography scan (CT) with the textbook “halo” sign that typically raises suspicion for Aspergillus in neutropenic hosts [56, 96]. Limitations notwithstanding, two retrospective studies do help establish an IPA prevalence range for patients with cirrhosis in the ICU, albeit a wide one. 84 patients in the smaller of the two studies had greater severity of both hepatic and overall organ failure. They were systematically evaluated for the presence of IPA using the galactomannan assay, and the prevalence was determined to be 14% [56]. In the larger study consisting of 986 patients, the severity of illness was lower, and potential cases were identified based on respiratory sample positivity for Aspergillus, which lacks sensitivity for IPA, with only a subset undergoing galactomannan testing [53]. In this scenario, the reported prevalence was 1.7%. Suggesting that even the upper boundary of this range might be an underestimate is an autopsy study that found hyphal invasion by Aspergillus in 15% of all ICU autopsies performed at a major referral center in Europe [97]. Viewed otherwise, among rigorously ascertained cases of IPA in the ICU, patients with liver disease have constituted 7–12% of all cases [98–100].

It is noteworthy that underlying cirrhosis has been shown to markedly increase the odds of Aspergillus spp. isolation on lower respiratory tract cultures in patients with critical influenza pneumonia [100]. This relates to the notion that the immune dysfunction of liver disease predisposes to greater fungal colonization, which then facilitates invasive infection under conducive conditions. A patient-level factor that may act synergistically with liver disease in promoting the conversion from Aspergillus colonization to invasion during critical illness is chronic obstructive pulmonary disease, a common comorbidity in patients with cirrhosis [56]. Another example of synergistic susceptibility to IPA occurs in cases of severe alcoholic hepatitis treated with corticosteroids. In fact, this is the scenario in which the link between liver disease and IPA was most apparent in the early years of its recognition. Interestingly, when IPA was studied systematically in a severe alcoholic hepatitis cohort of 92 patients, it was highly prevalent (15%), but corticosteroid exposure was not significantly higher in those who developed IPA [96]. In the same study, IPA was diagnosed almost invariably in the ICU setting and led to 100% mortality. Universal mortality was likewise observed in a study of critically ill cirrhotic patients with IPA [53]; even when non-universal in such a population, it is exceedingly high: 71% in a different study [56]. In a critically ill patient, the combination of liver disease and IPA clearly portends a very poor prognosis, though in a study limited to an ICU population with IPA, liver disease did not emerge as a risk factor for death [98]. A possible contributing factor to these dismal outcomes is the potential of voriconazole, the traditional second-generation triazole used for the treatment of IPA, to cause drug-induced liver injury, making it a problematic option in patients with baseline liver function impairment. Growing clinical experience with isavuconazole, the newest FDA-approved member of the triazole family, suggests a low rate of hepatic enzyme elevation attributable to this agent [101], which should facilitate the contemporary management of IPA in the setting of liver disease and hopefully improve its outcomes.

Cryptococcosis

Cryptococcus is a species of encapsulated yeast acquired by humans via the inhalational route classically following exposure to soil contaminated with pigeon droppings. Infection can remain localized to the lung in the form of nodules or consolidation or it can hematogenously disseminate to extrapulmonary sites, such as the central nervous system and the peritoneal cavity, usually in the setting of immunosuppression. Cryptococcosis gained prominence with the advent of the human immunodeficiency virus (HIV) as a prevalent cause of meningitis in patients with the acquired immunodeficiency syndrome (AIDS). As the evolution of antiretroviral therapy reduced the incidence of cryptococcal meningitis complicating AIDS in high-income countries, other risk factors began to receive greater recognition, among them liver disease. While cryptococcosis is not a diagnosis highly associated with an ICU stay in patients with cirrhosis, infection can be serious enough to warrant ICU admission. Examples of such scenarios include fungemia with shock, SFP, and cryptococcal meningitis, which could be misdiagnosed as refractory hepatic encephalopathy [102].

A contemporary prospective multicenter study of predominantly pulmonary cryptococcosis in patients without HIV found underlying liver disease in about 10% of the population [103]. On a national level, a US hospital discharge database study reported that the prevalence of liver cirrhosis and liver failure among those hospitalized for cryptococcal meningitis was 3–6 times higher, respectively, than that among all hospitalized persons during the same period [104]. The presence of liver disease appears to be enriched when cohorts of non-HIV patients with disseminated cryptococcosis (e.g., meningitis, fungemia) are considered separately. In such cohorts, liver cirrhosis was either the most or second-most frequently identified risk factor: when second, it is preceded only by receipt of immunosuppression, whether in the setting of organ transplantation or autoimmune disease [105–109]. In these cohorts, cirrhosis repeatedly emerged as an independent risk factor for disseminated cryptococcosis (aOR 5.3–8.5), especially cryptococcemia (aOR 23.8), on par with receipt of corticosteroids and other medications that suppress cell-mediated immune responses [108, 109]. Furthermore, one study identified cirrhosis as an independent risk factor for mortality in disseminated cryptococcosis and reported universal mortality in patients with cirrhosis [107]. When reported elsewhere, short-term (i.e., ≤ 90 days) mortality rates for this patient population have likewise been extremely high: in excess of 60% [106, 109–112] with the exception of a study in which ten-week mortality was quoted as 16.3% [113]. This discrepant result could be a function of the significant proportion of patients with Child–Pugh A cirrhosis (35%) in the study with the outlier survival figure.

In patients with liver disease, cryptococcosis often occurs in the presence of additional predisposing factors, such as diabetes mellitus and receipt of corticosteroids [112, 113]. As mentioned, infection is usually disseminated and thus antigenemia is common at the time of diagnosis though an attempt at screening asymptomatic cirrhotic patients with serum cryptococcal antigen testing did not detect any subclinical cases [114]. While the familiar pulmonary and central nervous system manifestations of cryptococcosis are also common in patients with cirrhosis, a noteworthy and long-recognized complication in these patients is SFP [115]. It occurs in approximately 20% of those with liver disease and cryptococcosis and raises intriguing questions about its pathogenesis, which is not well-understood [112]. Unlike Candida, the commonest cause of SFP, Cryptococcus does not normally inhabit the gastrointestinal tract, so the classical gut translocation paradigm underlying spontaneous peritonitis in patients with liver disease cannot be easily invoked. The portal of entry of Cryptococcus is the respiratory tract, from where the organism spreads hematogenously to remote sites in susceptible individuals. Though unproven, it would be reasonable to hypothesize that cryptococcal SFP develops as a result of seeding via the bloodstream in the setting of impaired fungal containment due to the global immune dysfunction of liver disease.

Conclusions and future perspectives

Patients with liver disease present many characteristic challenges in the ICU environment. Among them is their unique susceptibility to infection in the absence of traditional risk factors of such susceptibility like neutropenia or receipt of immunosuppressive medications. Akin to the coagulopathy of liver disease, its associated immunocompromised state is multifactorial, rendering liver disease patients vulnerable to both bacterial and fungal infections. It is important, therefore, for the ICU clinician to be familiar with the immune functions of the liver and the myriad ways in which hepatic and immunological impairment are linked. In the present review, we have summarized the current knowledge about defects in antifungal immunity that accompany liver disease, with many important scientific contributions to this field having been published in recent years. However, much remains unknown, especially surrounding mechanisms of systemic immunosuppression that place patients with cirrhosis at increased risk not only for infections acquired via the gastrointestinal tract but also via the respiratory route, such as IPA and cryptococcosis. As is likely the case with ICU patients in general, IFIs in the liver disease patient population remain underrecognized. Part of the reason is lagging adoption of sensitive molecular tests with rapid turnaround times, such as the magnetic resonance-based T2Candida™ Panel and the sandwich immunochromatography-based lateral flow assay for IPA. Unfortunately, even the most sensitive diagnostic tools available at the point of care will not help if the diagnosis of IFIs is not suspected by the clinical team. One of the present review's objectives is to raise awareness in the intensive care community about the frequent occurrence of IFIs in liver disease patients admitted to the ICU. Another is to highlight the extremely high associated mortality, which could at least in part be due to delayed recognition of IFIs in this fragile patient cohort. Although there has been growing acknowledgment by expert groups that hepatic dysfunction constitutes a predisposing host factor for IFIs, uptake of this concept in clinical practice has been very incremental. In the case of Cryptococcus, it took the decline of AIDS as the dominant risk factor for disseminated infection to pave the way for greater recognition of patients with cirrhosis as an at-risk population. Owing to the availability of effective—and now also minimally hepatotoxic—antifungal agents, the combination of heightened clinical suspicion and broad access to rapid testing could hold the key to reducing the unacceptably high ICU mortality of IFIs in liver disease. Finally, when it comes to prevention, the utility of antifungal prophylaxis in the ICU remains a controversial topic, and in this regard patients with liver disease have not been specifically studied.

Author contributions

OE conceived the article, reviewed and appraised the literature, drafted the manuscript, and approves the final version. AGM edited and critically revised the manuscript and approves the final version. JCO edited and critically revised the manuscript, assisted with illustration acquisition, and approves the final version. MSL edited and critically revised the manuscript and approves the final version.

Funding

This work was supported in part by the Division of Intramural Research, National Institute of Allergy & Infectious Diseases, NIH.

Declarations

Conflicts of interest

All authors declare that they have no relevant conflicts of interest.