Introduction

Fungi have emerged worldwide as an increasingly frequent cause of opportunistic infections. A survey of the epidemiology of sepsis conducted in the USA [1] revealed that the incidence of fungal sepsis increased threefold between 1979 and 2000. In contrast, numerous studies have revealed either no increase or sometimes even a decrease in the incidence of Candida sepsis [2–4]. Candida and Aspergillus spp. are the most frequent causes of invasive fungal infections and are associated with high morbidity and mortality [3, 5, 6]. The incidence of invasive candidiasis is sevenfold to 15-fold higher than that of invasive aspergillosis [3]. Originally described in immunocompromised hosts, primarily cancer patients, opportunistic fungal pathogens have now been recognized as a frequent cause of infection in surgical and critically ill patients.

The epidemiology of invasive mold infections is changing. Invasive aspergillosis is now also occurring in intensive care unit (ICU) patients, including mechanically ventilated patients and patients with chronic lung diseases treated with corticosteroids [7]. Moreover, the number of strains of non-fumigatus Aspergillus spp. is on the rise and multi-resistant non-Aspergillus mould infections are emerging. Although these are undoubtedly important epidemiological changes, this review article focuses on recent advances in our understanding of the epidemiology, diagnosis and treatment of invasive candidiasis, which is the predominant fungal infection occurring in critically ill patients.

Epidemiology

Candida is now the fourth leading micro-organism responsible for bloodstream infections in the USA, outnumbering all Gram-negative bacilli [8–10]. Data from 790 ICUs reporting to the US National Nosocomial Infection Surveillance system between 1990 and 1999 [8, 11] showed that Candida spp. were responsible for 5% to 10% of all bloodstream infections.

Studies of Candida infections in Europe have revealed significant differences from recent trends observed in the USA. In Europe, Candida is usually the sixth to the 10th cause of nosocomial bloodstream infections [4, 12–14]. In a survey conducted by the Fungal Infection Network of Switzerland between 1991 and 2000 [4], ICUs and surgical wards accounted for about two-thirds of all episodes of candidaemia. The incidence of candidaemia (on average 0.5 episodes/10,000 patient-days per year) was stable over this 10-year period and was five to 10 times higher in ICUs than in other wards.

During recent decades, several countries around the world have witnessed a change in the epidemiology of Candida infections, characterized by a progressive shift from a predominance of Candida albicans toward a predominance of non-albicans Candida spp. (including C. glabrata and C. krusei) [15]. C. glabrata has progressively increased and now accounts for 15% to 20% of infections in most countries [16, 17]. There is growing evidence suggesting a role in this epidemiological shift for increasing use of azole agents. Reduced susceptibility to commonly used antifungal agents has also been observed in some North American and European centres [18].

In ICU patients, the most common types of Candida infections are bloodstream infections, catheter-related infections, intra-abdominal infections and urinary tract infections [19–23]. Invasive candidiasis is recognized as a leading cause of morbidity and mortality in both immunocompetent and immunocompromised critically ill patients, with reported crude and attributable mortality rates of more than 40% to 60% and 20% to 40%, respectively [13, 23–29]. Of note, however, is that in the most recent clinical trials of new antifungal agents [30–35] the overall short-term (end of therapy) and long-term mortality (end of follow up) associated with candidaemia were found to be in the range of 15% to 20% and 30% to 40%, respectively (Figure 1). Candidaemia is also associated with prolonged duration of mechanical ventilation and hospital stay, and increased health care costs [28, 36–38].

Figure 1
figure 1

Mortality rates associated with Candida infections. Shown are rates of all-cause mortality from candidaemia or invasive candidiasis at (a) the end of antifungal therapy and (b) the end of follow up in recent randomized clinical trials. Numbers given in parenthesis on the x-axis indicate the reference numbers of the clinical trials. Duration of follow-up in panel a: 12 weeks, 22 to 4 weeks and 32 to 3 weeks. Duration of follow-up in panel b: 18 to 10 weeks and 212 to 14 weeks. AmB-d, amphotericin B-deoxycholate; Anidula, anidulafungin; Caspo, caspofungin; Flu, fluconazole; L-AmB, amphotericin B, liposomal preparation; Mica, micafungin; Vori, voriconazole.

Full size image

Risk factors

Two main factors predispose to infections with Candida spp.: colonization of skin and mucous membranes with Candida and alteration of natural host barriers (wounds, surgery, and insertion of indwelling intravascular and urinary catheters). The gastrointestinal tract, the skin and the urogenital tract are the main portals of entry for Candida infections. Colonization by Candida spp. has clearly been established as a major risk factor for invasive candidiasis [39]. Together with colonization with Candida induced by profound alteration of the endogenous flora resulting from prolonged broad-spectrum antibiotic therapy and loss of integrity of skin and mucosal barriers, surgery (especially of the abdominal compartment), total parenteral nutrition, acute renal failure, haemodialysis and treatment with immunosuppressive agents are major risk factors for invasive infections with Candida spp. [23, 25, 40]. Debilitating underlying diseases, critically ill status (as reflected by high Acute Physiology and Chronic Health Evaluation [APACHE] II score), antacids and mechanical ventilation have also frequently been associated with invasive candidiasis. Length of stay in the ICU is also associated with increased risk for Candida infections, which rises rapidly after 7 to 10 days [23, 29, 41, 43].

Prediction rules and scores for identification of non-neutropenic critically ill patients at risk for invasive candidiasis have been reported [39, 44–48]. Growth of Candida in semi-quantitative cultures (plating of specimens using the clock-streak technique and a calibrated loop) from multiple body sites has been used to predict the risk for invasive candidiasis [39]. The colonization index, calculated by dividing the number of colonized sites by the number of cultured sites, was found to be significantly higher in patients who developed invasive candidiasis than in control individuals (0.70 ± 0.17 versus 0.47 ± 0.17; P < 0.01) [39]. More recently, based on a prospective, cohort, observational, multicentre study that included 73 medical-surgical ICUs in Spain [48], a 'Candida score' was developed with the aim being to initiate antifungal therapy early. An adjusted logit model indicated that surgery on ICU admission, total parenteral nutrition, colonization at multiple sites with Candida and severe sepsis were associated with an increased risk for proven Candida infection. Patients with a Candida score, calculated using these variables, of 2.5 or more were 7.5 times more likely to have Candida infections than patients with a score of less than 2.5.

Most recently, an analysis of risk factors in 2,890 patients who stayed in the ICU for more than 4 days led to the development and validation of a clinical prediction rule for the early diagnosis of invasive candidiasis in the ICU [47]. The best prediction rule used a combination of the following factors: any systemic antibiotic or presence of central venous catheter and at least two other risk factors, including total parenteral nutrition, major surgery, pancreatitis, any use of steroids and use of immunosuppressive agents. This prediction rule exhibited a sensitivity of 34%, a specificity of 90%, a positive predictive value of 10% and a negative predictive value of 97%. This clinical rule may therefore help clinicians to rule out invasive candididiasis. However, data on the use of these risk assessment scores for guiding patient management are not yet available and their clinical utility remains to be established in prospective clinical studies.

Diagnosis

Given that rapid initiation of appropriate antifungal therapy is crucial for reducing mortality [13, 49], prompt diagnosis of infection is of the utmost importance. Unfortunately, diagnosing invasive fungal infections remains difficult and is often delayed. Indeed, blood cultures lack sensitivity (reported to be <50%) [50] and usually become positive late [51]. Invasive tissue sampling is often problematic in critically ill ICU patients. Radiological signs appear often late in the course of infection. Moreover, the European Organization for Research and Treatment of Cancer/Mycoses Study Group criteria for diagnosis of invasive mycoses [52], which are based on clinical, microbiological and radiological criteria, were developed in immunocompromised patients and may not apply to ICU patients. The need for sensitive and specific diagnostic tools has led investigators to look for non-culture-based methods aimed at detecting circulating fungal metabolites, antigens, antibodies and fungal DNA.

Serological tests consist of detection of components of the fungal cell wall, such as mannan, galactomannan and β-(1,3)-D-glucan, or antibodies directed against these antigens (antimannan) in blood or other body fluids. These tests have been shown to perform well in clinical studies. For example, three studies were conducted including 5% to 30% of critically ill patients [53–55]. Measurements of mannan and/or antimannan led to earlier diagnosis of Candida infection when compared with blood cultures [53, 54]. Sensitivity and specificity (respectively) were 40% and 98% for mannan and 53% and 94% for anti-mannan antibodies, and 80% to 90% when combining the two tests [55]. Assays for detection of β-(1,3)-D-glucan are used widely in Japan, and one of these assays (Fungitell; ACC, Falmouth, MA, USA) was recently approved by the US Food and Drug Administration. Studies conducted with β-(1,3)-D-glucan assays have yielded sensitivities ranging from 69% to 97%, specificities ranging from 87% to 100%, and positive and negative predictive values ranging from 59% to 96% and 75% to 97%, respectively [56–59]. Given these excellent negative predictive values β-(1,3)-D-glucan tests can help to rule out invasive candidiasis. Unfortunately, little information has been published thus far on use of β-(1,3)-D-glucan tests in the ICU setting.

Molecular diagnostic tests for detection of Candida DNA in either blood or tissues have been described [60, 61]. Albeit promising, relatively few data have been published on the performance of the detection of fungal DNA in high-risk critically ill patients. In addition, these tests are not yet commercially available.

Noninvasive diagnostic tools look promising for early diagnosis of invasive candidiasis. Clinical studies should now be conducted to evaluate their utility for guiding therapeutic decisions (see Pre-emptive therapy, below).

Antifungal therapy

Prophylaxis

Few prophylactic studies have been performed in ICU patients [43, 62–67]. Earlier studies conducted by Savino and coworkers [64] and Slotman and Burchard [63] compared the efficacy of prophylactic administration of oral clotrimazole, ketoconazole, or nystatin with that of placebo in patients selected based either on expected length of stay in the ICU or on baseline risk factors. The results of these under-powered studies revealed either no effect or only a modest impact of prophylaxis on occurrence of Candida infections [68].

In contrast, several more recent studies [43, 62, 65] indicated that high-risk critically ill patients may benefit from antifungal prophylaxis. Fluconazole prophylaxis was found to prevent intra-abdominal candidiasis in high-risk surgical patients with recurrent gastrointestinal perforations or anastomotic leaks [65]. The risk for intra-abdominal candidiasis was reduced eightfold in patients receiving fluconazole (400 mg/day). One fluconazole-treated patient (4%) developed Candida peritonitis as compared with seven placebo-treated patients (35%; P = 0.02). The number of patients needed to prevent one episode of intra-abdominal candidiasis was 3, indicating that prophylaxis had considerable impact. Four (20%) patients died from fungal infections in the placebo group, but none did so in the fluconazole group (P = 0.04). In a randomized, double-blind, placebo-controlled trial conducted in medical and surgical ICU patients ventilated for at least 48 hours and expected to stay in the ICU for another 72 hours [62], fluconazole prophylaxis (100 mg/day) exerted a modest protective effect against Candida colonization. Although it did not prevent the development of severe Candida infections, which was the primary study end-point, fluconazole prophylaxis markedly reduced the number of episodes of candidaemia. In the third study, that conducted by Pelz and coworkers [43] in 260 surgical patients expected to stay in the ICU for more than 3 days, 11 (9%) fungal infections occurred in the fluconazole group as compared with 20 (16%) in the placebo group (P < 0.05). Mortality was similar between the two treatment groups.

Overall, these three classic studies strongly suggest that azole prophylaxis has the capacity to reduce the incidence of invasive candidiasis in surgical and ICU patients. However, an important issue remains how to identify those patients who are likely to benefit from prophylaxis without unnecessarily exposing patients who are at either low or no risk to antifungal agents. Indeed, according to a Cochrane review on antifungal agents for the prevention of fungal infections in non-neutropenic critically ill patients [69], the number of patients who should be treated with fluconazole to prevent one Candida infection is 94. This estimate, based on an incidence of fungal infection of 2%, ranged from 9 in high-risk patients to 188 in low-risk patients. Whether antifungal prophylaxis may have an impact on mortality remains a matter of debate. Although no individual study demonstrates an impact of azole prophylaxis on mortality, the recent Cochrane meta-analysis [69] indicated that prophylaxis did reduce the overall mortality in non-neutropenic critically ill patients. In the 2004 guidelines of the Infectious Diseases Society of America on treatment of candidiasis [19], routine use of antifungal prophylaxis in the general ICU setting was discouraged. However, it was suggested that fluconazole prophylaxis should be considered in carefully selected patients (a recommendation classified as A1, based on the strength of the evidence). These guidelines are being revised and an updated version should be available in 2008.

Pre-emptive therapy

There is an extreme paucity of studies on pre-emptive antifungal therapy. In a study conducted between 1998 and 2002 in a surgical ICU in France [70], administration of targeted pre-emptive intravenous fluconazole therapy (fluconazole: 800 mg loading dose and then 400 mg/day for 2 weeks) based on colonization indexes was shown to prevent development of proven candidiasis in ICU patients, when compared with an historical control group of patients. A study conducted in Japan examined the effects of early initiation of pre-emptive therapy with an azole (fluconazole or miconazole in 78% and 2% of patients, respectively) or an echinocandin (micafungin in 20%), which was initiated based on a combination of Candida colonization at multiple sites and a positive β-(1,3)-D-glucan test [71]. The findings indicated that early pre-emptive strategy prevented candidaemia but had no impact on mortality.

Treatment of documented Candida infections

Polyenes

For decades amphotericin B deoxycholate has been the standard therapy for invasive fungal infections. Unfortunately, amphotericin B deoxycholate is often poorly tolerated and associated with acute infusion-related reactions and nephrotoxicity. During the late 1970s and 1980s, the development of azoles (miconazole, ketoconazole, fluconazole and itraconazole) provided alternative therapeutic options to amphotericin B for the treatment of candidiasis. In recent years, several new antifungal agents have become available, further enlarging the antifungal armamentarium (Table 1) [30–35]. These include lipid formulations (colloidal dispersion, lipid complex and liposomal) of amphotericin B, new azoles (voriconazole and posaconazole) and echinocandins (caspofungin, micafungin and anidulafungin). Lipid formulations of amphotericin B (colloidal dispersion, lipid complex and liposomal) are better tolerated than amphotericin B deoxycholate and have been used mainly in patients who are intolerant to conventional amphotericin B or are unlikely to tolerate it because of altered renal function. Few studies have compared the efficacy of amphotericin B deoxycholate with that of lipid formulations for the treatment of patients with invasive candidiasis [72, 73]. Small noncomparative studies [72, 73] suggested that lipid formulations of amphotericin B are as efficacious as conventional amphotericin B. High costs, a relative paucity of clinical data and existence of alternative antifungal therapies (azoles and echinocandins) explain why lipid formulations have generally been used as second-line therapy in patients with refractory invasive candidiasis.

Table 1 Randomized multicentre clinical trials of antifungal therapy in patients with candidaemia or invasive candidiasis
Full size table

Triazoles

In a multicentre study in non-neutropenic patients with candidaemia, fluconazole (400 mg/day) was found to be as efficacious as and better tolerated than amphotericin B deoxycholate (0.5 to 0.6 mg/kg per day) [31]. Fluconazole remains one of the most commonly used antifungal agents for the treatment of Candida infections. However, innate (C. krusei) or emerging (especially C. glabrata and C. guilliermondi) resistance to azoles among non-albicans Candida spp. has been noted in various regions of the world [16, 17], which may limit the use of fluconazole as empirical therapy for yeast bloodstream infections in critically ill patients before species identification and results of antifungal susceptibility testing are known. Data on the efficacy of high doses (800 to 1,200 mg) of fluconazole for treatment of less susceptible Candida strains are lacking.

Voriconazole, a second-generation triazole that is active against all Candida spp., is a new option for intravenous and oral therapy of Candida infections [74]. In a randomized, open-label, comparative multicentre, noninferiority trial conducted in patients with invasive Candida infections [33], voriconazole (6 mg/kg per day after a 12 mg/kg loading dose on day 1) was shown to be at least as effective as and safer than amphotericin B deoxycholate (0.7 to 1 mg/kg per day) followed by intravenous or oral fluconazole (400 mg/day). Transient, fully reversible visual adverse events and abnormalities of liver function tests are observed in 20% to 40% and 5% to 15% of patients treated with voriconazole, respectively. Efficacy of and/or tolerance to voriconazole may be affected by great variability in blood levels caused by nonlinear pharmacokinetics, polymorphism of cytochrome CYP2C19, drug-drug interactions and hepatic dysfunction [75–77]. Monitoring of circulating drug concentrations to target trough blood values between 1–2 and 6 mg/l would appear prudent, especially during the acute phase of life-threatening infections [78, 79].

Itraconazole (an azole that may be admininstered by oral and intravenous routes) and posaconazole (a new oral azole with a broad spectrum of antifungal activity against Candida spp., Aspergillus spp. and other emerging molds, including Fusarium spp. and zygomycetes) have been shown to be efficacious for treatment of oropharyngeal candidiasis [80, 81]. However, no comparative clinical trials in patients with candidaemia have been performed with these antifungal agents, and their efficacy in this clinical setting remains to be determined. One concern, however, might be the potential risk for development of cross-resistance, which could limit the utility of new azoles for therapy of infections due to non-albicans Candida spp.

Echinocandins

Echinocandins are a new class of parenteral antifungal agents that inhibit the synthesis of β-(1,3)-D-glucan in the fungal cell wall [82]. These compounds are fungicidal in vitro against C. albicans and non-albicans Candida spp. No cross-resistance with azoles has yet been reported. Three agents are available for clinical use [42, 83]: caspofungin, micafungin and anidulafungin. The safety profile of echinocandins is excellent, with few reported adverse events (abnormal liver function tests, phlebitis, or histamine-like reactions). Drug-drug interactions with some medications have been observed with caspofungin (for example, with rifampicin, anticonvulsants, tacrolimus, cyclosporin, protease inhibitors and non-nucleoside reverse transcriptase inhibitors).

Caspofungin was the first echinocandin to be licensed for the treatment of invasive mycoses, including candidiasis [82]. In immunocompromised (mainly HIV-positive) patients with oropharyngeal and/or oesophageal candidiasis, caspofungin was found to be as effective as amphotericin B deoxycholate or fluconazole [84–86]. In a multicentre trial conducted inpatients with invasive candidiasis, caspofungin (50 mg/day after a 70 mg loading dose) was at least as efficacious as and less toxic than amphotericin B deoxycholate (0.6 to 1 mg/kg per day) [32]. Recent reports have described the emergence of resistance to caspofungin in patients with oesophagitis, candidaemia and endocarditis [3]. In a multicentre, randomized, double-blind trial, micafungin (100 mg/day) was as effective as and less toxic than liposomal amphotericin B (3 mg/kg per day) for first-line therapy of candidaemia or invasive candidiasis [34]. In a randomized, double-blind study conducted in patients with invasive candidiasis [35], anidulafungin (100 mg/day after a 200 mg loading dose) was observed to be superior to fluconazole (400 mg/day after a 800 mg loading dose), but the study was reported to show noninferiority after removal of the centre that enrolled the largest number of patients. A recent, randomized, double-blind study comparing micafungin (100 or 150 mg/day) and caspofungin (70 mg loading dose and then 50 mg/day) in 595 adult patients with candidaemia or invasive candidiasis [30] reported noninferior efficacy of micafungin compared with that of caspofungin and similar safety profiles for the two compounds.

Thus, recent studies have shown that echinocandins are efficacious and safe, explaining why this new class of antifungal agents has assumed a prominent role in the management of patients with invasive candidiasis.

Combinations of antifungal agents

Given the poor prognosis of Candida sepsis in critically ill patients, clinicians have shown interest in using combinations of antifungal agents of different classes. Amphotericin B deoxycholate and 5-flucytosine have been shown to be synergistic in vitro and in experimental models of candidiasis [87–89]. Combination of fluconazole and amphotericin B has been shown to be antagonistic in experimental models of aspergillosis, but not in models of invasive candidiasis [90, 91]. However, there is a dearth of information available from few clinical studies. In a randomized, double-blind study conducted in non-neutropenic patients with candidaemia [92], high-dose fluconazole (800 mg/day intravenously) was compared with a combination of fluconazole (800 mg/day intravenously) and amphotericin B deoxycholate (0.7 mg/kg per day intravenously). At first glance, the efficacy of combination therapy was slightly superior to that of monotherapy (success: 69% versus 56%), especially in patients with an APACHE II score ranging between 10 and 22. However, there were statistically significant differences in baseline covariates between the two groups, such as APACHE II score, which was lower in the combination treatment arm. Until clinical trials are reported that demonstrate efficacy and safety, the indiscriminate use of combination therapy in patients with invasive candidiasis should be discouraged.

Conclusion

Invasive candidiasis is the most frequent invasive mycosis in critically ill patients. Changing epidemiology with increased non-albicans Candida spp., nonspecific risk factors and clinical presentation, and late diagnosis with culture-based methods are major challenges in the management of invasive candidiasis. Preventive strategies targeting patients with a high-risk profile, development of new noninvasive diagnostic tools that allow early diagnosis and therapy, and extension of the therapeutic armamentarium with new agents are encouraging recent advances that may allow us to overcome Candida infections.

Note

This article is part of a review series on Infection, edited by Steven Opal.

Other articles in the series can be found online at http://ccforum.com/articles/theme-series.asp?series=CC_Infection