Diagnosis and management of invasive candidiasis in critically ill patients: SIAARTI multidisciplinary statement

Most critically ill patients admitted to ICUs are at risk of developing invasive Candida infections [2, 3, 9]. This is supported by the presence of largely non-specific risk factors related to comorbidities, treatments, consequences of therapies or underlying disease, and/or the devices required for organ support. A recent meta-analysis identified 29 risk factors from 34 studies, encompassing demographic factors, treatments received during hospitalisation, and comorbidities [2]. Among these, neutropenia, HIV, and multisite Candida colonization had the most significant impact. Regarding therapeutic interventions, the presence of a central venous catheter (OR 4.7; 95% CI 2.7–8.1); broad-spectrum antibiotic therapy (OR 5.6; 95% CI 3.6–8.8); transfusions (OR 4.9; 95% CI 1.5–16.3); total parenteral nutrition (OR 4.6; 95% CI 3.3–6.3); and Candida colonization (OR 4.7; 95% CI 1.6–14.3) were the strongest associations in the final statistical model. Interestingly, demographic factors were not significantly associated with the risk of Candida infections. Furthermore, significant collinearity between risk factors highlighted their interdependence, making it difficult to isolate the true independent association of each factor, especially in critically ill patients.

Candida spp. colonization is a particularly critical risk factor that requires special attention because it represents a marker of the interaction between the host and the microorganisms. A recent meta-analysis found that ICU patients with Candida colonization have a significantly increased risk of developing invasive candidiasis (OR 3.32; 95% CI 1.68–6.58) compared to non-colonized patients [10].

Another recent meta-analysis found that the average ICU stay at the onset of candidemia was 12.9 days (95% CI 11.7–14.2) [9]. It is reasonable to consider that the cumulative impact of these risk factors increases with the length of ICU stay, as the pathophysiological process leading to invasive candidiasis has time to develop and progress (e.g., central venous catheter insertion, colonization, candidemia) [9, 11]. Regarding intra-abdominal candidiasis, a recent retrospective study including patients from 26 ICUs in European hospitals identified recurrent intestinal perforation (OR 13.90; 95% CI 2.65–72.82); anastomotic dehiscence (OR 6.61; 95% CI 1.98–21.99); the presence of abdominal drainage (OR 6.58; 95% CI 1.73–25.06); antifungal treatment (OR 4.26; 95% CI 1.04–17.46); or antibiotic treatment for seven or more days (OR 3.78; 95% CI 1.32–10.52) as independently associated with the development of intra-abdominal candidiasis [12].

The appropriateness of non-targeted antifungal strategies in critically ill patients—i.e. the administration of antifungal agents before a definitive microbiological diagnosis—is a long-debated topic. The use of empirical antifungal therapy in critically ill ICU patients is widely debated due to the lack of clear evidence supporting a survival benefit [13]. Specifically, the cost–benefit balance of initiating empirical antifungal therapy (i.e. treatment in patients presenting with non-specific signs and symptoms of infection despite antibiotic therapy) in critically ill patients remains controversial despite decades of research [13, 14].

Two major randomised controlled trials [15, 16] indicate that initiating empirical antifungal therapy alone does not improve survival. Moreover, there is currently no strong evidence demonstrating a consistent mortality reduction with early empirical therapy [13]. However, evidence is heterogenous in terms of patient populations, settings, necessity of source control, and overall management, thus an overall benefit cannot be excluded [17]. Thus, there is a discrepancy between the association between early initiation of therapy and mortality in observational studies [18‐20] and the findings of randomized clinical trials, and this underlines the difficult to define an appropriate overall strategy [21].

The 2021 Surviving Sepsis Campaign guidelines identified additional risk factors beyond those already described, including immunosuppression; neutropenia; clinical severity (e.g., high APACHE score); prior surgery; emergency gastrointestinal and hepatobiliary surgery; acute kidney failure; and dialysis [22]. These guidelines recommended that the initiation of empirical antifungal therapy should be based on risk factor assessment. Specifically, antifungal treatment is suggested for patients with sepsis or septic shock who are at high risk for fungal infections, while it is not recommended for those at low risk [22]. A task force from ESICM/ESCMID recommends initiating empirical antifungal therapy in patients with septic shock, multi-organ failure, and confirmed Candida colonization in at least one extra-digestive site (strong recommendation, low level of evidence) [23].

The latest clinical practice guidelines by the American Thoracic Society suggests against routine prophylactic or empiric antifungal agents against Candida spp. In critically ill patients without neutropenia or a history of transplant [24]. The recent global guidelines on Candida infections by the European Confederation of Medical Mycology (ECMM) and International Society of Human and Animal Mycology (ISHAM) moderately recommend starting empiric antifungal treatment for patients with septic shock or patients with deteriorating conditions with additional risk factors for candidemia, such as prolonged ICU stay, indwelling vascular catheter and Candida spp. colonization [4].

In this context, risk factor assessment, alongside the evaluation of clinical severity, is a key determinant in the decision to start empirical antifungal therapy in suspected invasive candidiasis, ensuring an appropriate evaluation of the cost–benefit ratio. A comprehensive case-by case clinical decision between initiation of empiric antifungal treatment versus watchful waiting and eventual timely initiation of targeted antifungal therapy seems to be the most reasonable approach to balance risks and benefit [25].

The beta-D-glucan (BDG) test has been proposed as a useful tool to guide the initiation of empirical antifungal therapy, as it detects components of the Candida spp. cell wall, indicating potential invasive infection. A study by Christner et al. [26] evaluated the effectiveness of BDG testing in guiding therapeutic decisions for high-risk, critically ill patients. The results showed that BDG testing has moderate sensitivity (74%) and low specificity (45%) when using standard thresholds. However, very high BDG levels improved the test’s predictive ability, suggesting that BDG may help identify patients who could benefit from early antifungal therapy. The multicentre CandiSep trial [27] further investigated the use of BDG testing to guide antifungal therapy in sepsis patients at high risk of invasive candidiasis. While BDG testing enabled earlier initiation of antifungal therapy (median of 1.1 days in the BDG group vs. 4.4 days in the control group, p < 0.01), it did not lead to a significant reduction in 28-day mortality (33.7% in the BDG group vs. 30.5% in the control group; p = 0.53). Additionally, BDG testing was not shown to be cost-effective and associated to over-treatment since a significantly higher proportion of patients in the BDG group received antifungal within 96 h versus the control group (48.8% vs. 6%). The Candi-NET study reaffirmed that no robust data directly correlate high BDG levels with an increased risk of developing the disease. On the other hand, at least two studies have confirmed the high negative predictive value of BDG testing, making it particularly useful for early discontinuation of antifungal therapy in critically ill patients, especially in cases of negative test results [28, 29]. This suggests that while BDG testing may accelerate treatment initiation, it does not necessarily improve clinical outcomes unless combined with other risk factors and clinical indicators and may carry the risk of over-treatment. The high negative predictive value of BDG is a key factor, and its use as a tool for early discontinuation of antifungal therapy could provide significant clinical advantages. Specifically, in patients with septic shock or suspected fungal infection, a negative BDG test can help reduce unnecessary treatments and minimise the risk of antifungal-related adverse events [30, 31]. Conversely, in ICU settings, a reduction in BDG serum concentration appears to correlate with lower mortality. Notably, a > 70% BDG reduction has been associated with increased survival, with a specificity and positive predictive value of 100% [32]. The ECMM/ISHAM Candida global guidelines for diagnosis and management of candidiasis reports that diagnosis of IC should not be solely based on BDG testing and positive serum BDG alone is not recommended for initiating antifungal treatment. Indeed they moderately support the use of BDG testing to discontinue empirical antifungal treatment [4].

The appropriateness of antifungal therapy, even in its empirical phase, is essential to optimize the treatment of invasive candidiasis in critically ill patients. Numerous observational studies have demonstrated that, in addition to controlling the infectious source, selecting the appropriate antifungal agent is a key factor in improving the prognosis of patients with severe infections [12]. It is well established that risk factors for invasive Candida infections are multiple and non-specific, including prolonged ICU hospitalisation, broad-spectrum antibiotic therapy, immunosuppression, and complex abdominal surgery [13]. As a result, most antifungal therapies for Candida infections are initiated in patients without a documented infection, contributing to widespread antifungal use and increasing ecological pressure. This is evidenced by the rising prevalence of fluconazole-resistant Candida non-albicans species [33]. For these reasons, current international guidelines recommend the use of echinocandins (caspofungin, anidulafungin, micafungin, or rezafungin) as first-line therapy in critically ill patients with suspected or confirmed invasive candidiasis, followed by L-AmB as second-line, particularly in patients who have already been treated or who have risk factors for Candida non-albicans species resistant to echinocandins [4, 23, 34]. Unlike azoles, which exhibit time-dependent pharmacological activity and a fungistatic in vitro effect, echinocandins and L-AmB share concentration-dependent activity and a potent fungicidal action, allowing for rapid fungal load reduction [35]. A literature review including seven randomised studies with 1915 patients with candidemia and invasive candidiasis showed that echinocandin therapy was associated with reduced mortality (OR 0.65; 95% CI 0.45–0.94) and a higher clinical cure rate (OR 2.33; 95% CI 1.27–4.35) [36]. Conversely, while amphotericin B has the highest fungicidal activity and a prolonged post-fungal effect, it is still considered a second-line option due to the high toxicity risk of its original formulation (amphotericin B deoxycholate). However, a literature review involving 2352 patients found no superiority of echinocandins over amphotericin B as a first-line treatment. Instead, the liposomal formulation of amphotericin B was associated with a significantly lower risk of nephrotoxicity [36]. In critically ill patients, azoles should note be considered first-line agents for IC. Their role is mainly limited to step-down once clinical stability is achieved and susceptibility to common azoles is confirmed. This approach is both ecologically responsible and clinically safe in terms of efficacy and adverse events [37]. In settings without routine access to susceptibility testing, local epidemiology and resistance pattern should be taken into account before considering de-escalation to azoles. The ECMM-ISHAM global guidelines on IC marginally recommend fluconazole and voriconazole as second-line treatment for candidemia for the high-risk of treatment failure, the increased antifungal resistance (i.e. azole-resistance C. parapsilosis, C. auris, C. glabrata), drug-to-drug interaction and the need for TDM [4]. The same guidelines moderately recommend switching to an oral azole (fluconazole or voriconazole) after 5 or more days of echinocandin treatment in patients with haemodynamic stability, documented clearance of Candida from the bloodstream, absence of neutropenia, completed source control, ability to tolerate oral azole treatment and susceptibility confirmed to selected azoles [4]. For infections involving selected anatomical sites (e.g. CNS or ocular infections) their favourable tissue penetration may make them a suitable therapeutic option.

Another critical factor in treating invasive candidiasis is the site of infection and the tissue concentration profile of different antifungal agents. In patients with primary candidemia or Candida-associated infections where the device has been removed, the pharmacokinetic and pharmacodynamic (PK/PD) profile of echinocandins supports their role as first-line agents, given their high safety profile and reliable achievement of PK/PD targets, even in patients with septic shock or multi-organ failure.

However, several anatomical sites remain difficult to penetrate because of the hydrophilic nature of echinocandins and their high plasma protein binding, which limit tissue distribution. Current guidelines recommend azoles or polyenes as the only therapeutic alternatives for infections involving the CNS, ocular compartment, pericardial/pleural serous cavities, osteoarticular system, and urinary tract, depending on disease severity, susceptibility profile of the isolates, and the risk of adverse events [4].

A particular consideration is intra-abdominal candidiasis, one of the most common forms of invasive Candida infection. In this setting, fluconazole remains an appropriate choice for clinically stable patients with fully susceptible fungal isolates. However, in critically ill patients or cases with a high risk of azole resistance, the use of echinocandins as first-line therapy has been questioned [37]. Emerging pharmacokinetic studies has shown that caspofungin, anidulafungin, and micafungin achieve only ~ 15–30% penetration rate into peritoneal fluid, raising concerns about treatment failure and resistance selection, particularly with prolonged therapy [38]. To address this, strategies include the following:

For device-associated infections (e.g. endocarditis, prosthetic infections, or synthetic material infections), the presence of biofilm limits the effectiveness of several antifungals. Echinocandin are fungicidal against planktonic Candida spp. and retain activity against sessile form in biofilms while lipososomal amphotericin B (L-AmB) is active against both planktonic and sessile form. Azoles generally have limited activity against biofilm. Consequently, when biofilm Is suspected, an echinocandin or L-AmB are appropriate options in addition to prompt device removal when feasible [8]. Finally, except in rare cases of profoundly immunosuppressed patients with histological evidence of tissue invasion, the isolation of Candida spp. from respiratory secretions should be considered colonization and does not require antifungal treatment [39].

The emergence of resistant Candida species poses a major challenge in the management of invasive candidiasis in critically ill patients [4, 33]. Among these, C. auris has gained increasing relevance in ICU settings, being responsible for nosocomial outbreaks due to its ability to persist in the environment, its high transmission potential, and its frequent multidrug-resistant profile [40, 41]. Echinocandins are generally considered the first-line agents when C. auris is suspected or confirmed, although reduced susceptibility and resistance have also been described [4]. In addition, fluconazole resistance has become a growing concern in several countries, including in Europe. In a recent multicentre observational study, fluconazole resistance was found in 17% of C. parapsilosis isolates (from Italy, Turkey, and Greece) and 12% of C. glabrata isolates (from six European countries) [42]. This resistance significantly limits the use of fluconazole as a step-down agent and highlights the importance of local epidemiological data and susceptibility testing to guide antifungal therapy. These evolving resistance patterns reinforce the need for continuous surveillance and an individualized approach to antifungal selection in critically ill patients.

Echinocandins inhibit β-1,3-D-glucan synthesis, a key component of the fungal cell wall, and exert fungicidal activity against Candida spp. However, some species, such as C. parapsilosis, have higher minimum inhibitory concentrations (MICs) [43, 44]. The antifungal effect of echinocandins is concentration-dependent and characterised by a post-antifungal effect, making them the first-choice therapy for invasive candidiasis in critically ill patients [8]. First-generation echinocandins (caspofungin, micafungin, and anidulafungin) are administered parenterally once daily. The have high protein binding (> 95%) and good tissue distribution, particularly in the liver, kidneys, and lungs. However, ocular and CNS penetration is very limited, and urinary excretion is negligible [43]. Although hepatotoxicity and cardiotoxicity have been described with echinocandins in preclinical models [45, 46], the clinical impact of these findings is negligible. Conversely, the lack of clinically relevant drug-drug interactions may represent a significant advantage with the use of echinocandins [47]. Rezafungin, a new-generation echinocandin derived from anidulafungin, is characterised by an extended elimination half-life (> 130 h), allowing for once-weekly administration [41]. Phases II and III trials have demonstrated its non-inferiority to caspofungin in treating invasive candidiasis, leading to recent approval for this indication in Italy [48, 49].

Recommended dosing regimens for the different echinocandins are presented below:

L-AmB is an improved formulation of amphotericin B, encapsulated in a liposomal structure. This enhances its pharmacological properties since the liposoma acts as a sump while reducing the nephrotoxicity risk associated with older formulations. Its mechanism of action involves binding to fungal ergosterol with greater affinity than human cholesterol. In vitro, L-AmB exhibits fungicidal activity with a broad spectrum of action and a minimal risk of resistance development. L-AmB is administered exclusively intravenously in bound for ~ 95% to plasma proteins and reaches high tissue concentrations in the infected tissues thanks to the peculiar pharmacokinetic behaviour of the liposomal moiety [50]. Compared to conventional amphotericin B, L-AmB may be used at higher doses per kg without increasing the toxicity risk. Once absorbed, the reticuloendothelial system (RES) sequesters L-AmB, with the liver, kidneys, and lungs serving as primary reservoirs. L-AmB achieves low cerebrospinal fluid (CSF) penetration (~ 2–4% of serum levels), although this may reach up to 90% in children [51]. The standard dose (3 mg/kg/day) provides adequate intra-abdominal penetration, particularly in the presence of inflammation [45]. However, a recent trial suggested that a higher dose (5 mg/kg) may be helpful in preventing disease development in critically ill patients at high risk of intra-abdominal invasive candidiasis [52].

Azoles act by inhibiting ergosterol synthesis, a key component of the fungal cell membrane. Their activity is mainly fungistatic. Fluconazole is used for Candida infections, whereas triazoles extend their spectrum to include moulds but show variable activity against Candida spp. [53]. Azoles generally achieve excellent tissue penetration, though specific pharmacokinetic differences between agents may exist. For example:

The main toxicity concern with azoles is hepatic dysfunction, which can manifest as elevated cytolytic and/or cholestatic markers. Additional side effects include the following:

For these reasons, fluconazole is commonly used as step-down therapy, including oral administration, for candidemia and intra-abdominal invasive candidiasis (see statement rationale 3.1) [4].

Recommended dosing regimens for the different azoles are reported below:

Fluconazole is a hydrophilic drug primarily eliminated by the kidneys, and its plasma exposure can vary significantly in critically ill patients due to their specific pathophysiological conditions, potentially affecting the achievement of an optimal PK/PD target [36, 55]. Preclinical studies have identified the AUC/MIC (area under the curve/minimum inhibitory concentration) ratio as the best pharmacodynamic efficacy index for fluconazole. Specifically, an AUC/MIC ratio greater than 25–50 has been associated with maximal antifungal effect, even against Candida spp. expressing various resistance mechanisms [56]. A retrospective study of 77 critically ill, non-neutropenic patients with candidemia demonstrated that achieving an AUC/MIC ratio above 55.2 was associated with a significantly increased probability of survival (p = 0.008) [57]. In liver transplant patients, fluconazole penetration into the peritoneal and biliary compartments has been shown to be sufficient to reach the optimal PK/PD target [58]. Based on these findings, it has been proposed that implementing a TDM-guided strategy to maintain minimum plasma concentrations of fluconazole between 10–20 mg/L could maximise the achievement of an optimal PK/PD target in critically ill patients with abdominal-source ICI or post-liver transplantation [38, 58, 59]. This strategy may be particularly beneficial in high-risk subpopulations prone to underexposure with standard doses, such as patients with morbid obesity, augmented renal clearance, or those undergoing continuous renal replacement therapy (CRRT) [60, 61].

Echinocandins are recommended as first-line therapy for abdominal-source IC according to international guidelines [4] However, their role in this setting has recently been debated following new evidence demonstrating poor penetration into peritoneal fluid, which increases the risk of suboptimal drug exposure and resistance selection, particularly in critically ill patients [62]. Caspofungin, micafungin, and anidulafungin exert a concentration-dependent antifungal effect, and the fAUC/MIC ratio has been identified as the best pharmacodynamic efficacy index in both preclinical and clinical studies (> 20 for C. albicans, > 7 for C. glabrata and C. parapsilosis). Clinical studies have shown that peak concentrations and AUC values for all the three echinocandins can be significantly lower in critically ill compared to non-critically ill patients and to healthy volunteers [63, 64]. Other studies have demonstrated that in patients with abdominal-source IC, the penetration of caspofungin, anidulafungin, and micafungin into ascitic fluid is limited, preventing the achievement of optimal PK/PD targets at the infection site and potentially favouring the selection of resistant strains [65‐67]. As a result, several studies have highlighted the need for higher loading and maintenance doses of echinocandins in critically ill patients with abdominal-source ICI [68, 69]. Where TDM is unavailable, increasing standard doses should be considered an alternative approach to maximise the likelihood of achieving effective PK/PD targets, thereby reducing the risk of clinical failure and resistance selection. This strategy is particularly important in critically ill patients with physiological limitations that impair drug absorption or tissue distribution, such as those with peritoneal infections or septic shock [70].

L-AmB exhibits distinct PK/PD characteristics due to its unique pharmaceutical formulation, setting it apart from fluconazole and echinocandins in managing critically ill patients with abdominal-source IC. Its total plasma concentrations are significantly higher than those observed with amphotericin B deoxycholate, even when doses are normalised per kilogram of body weight. It is believed that most circulating drug is biologically inactive, with the liposomal formulation acting as a reservoir, releasing active drugs upon direct contact with the membrane of the fungus [71]. Unlike other antifungals, L-AmB has a predominantly non-renal and non-hepatic clearance, along with a prolonged elimination half-life, making it less susceptible to pathophysiological alterations commonly observed in critically ill patients [56, 62]. Preclinical studies have suggested that achieving a Cmax/MIC ratio > 10 may be associated with improved fungicidal activity. However, clinical evidence linking this PK/PD target to better efficacy remains limited [56]. A recent systematic review found no sufficient evidence to support the use of a TDM-guided strategy with Cmax/MIC targets of 25–50 for improving efficacy or AUC24h < 600 mg·h/L for minimising nephrotoxicity risk [72]. Consequently, implementing a TDM-guided strategy for L-AmB is currently neither recommended nor considered clinically useful.

This document has several limitations that should be acknowledged. First, it is not a GRADE-based document; therefore, clinicians should primarily rely on evidence-based clinical practice guidelines for decision-making on this topic. Second, no formal systematic review of the literature was performed, which may have introduced potential bias in the selection of supporting evidence. Third, the multidisciplinary nature of the panel may have broadened the perspective, which could sometimes dilute the specific focus on critically ill patients. Nonetheless, the inclusion of panelists with long-standing experience in this setting was intended to maintain the applicability of the recommendations to critically ill patients.