Despite the significant progress in the supportive and intensive care of patients at risk, the development of a number of new and useful antifungal agents in the past decade, and the noteworthy improvements in therapeutic approaches to fungal infections, physicians' ability to diagnose these infections in a timely fashion remains limited, and patient outcomes remain poor. Antifungal prophylaxis has emerged as a potential means of reducing the occurrence of serious fungal infections. In patient populations estimated to be at high risk for acquiring a fungal infection, antifungal prophylaxis has reduced infection rates by about 50%; however, it has not been shown to improve mortality significantly.

This chapter outlines an approach to the workup and management of the nonneutropenic surgical patient with a suspected candidal infection, taking into account epidemiology, risk factors, diagnostic evaluation, and therapy. Noncandidal infections (aspergillosis and zygomycosis) are considered as well.

Fungal pathogens can be isolated under a number of clinical conditions.^2–7 Isolation of a fungal pathogen in the absence of any signs or symptoms of clinical infection is termed colonization. Fungal colonization, by itself, does not suffice to define fungal infection, but it remains an important risk factor for future invasive fungal infection. It is believed that once colonization has occurred, Candida pathogens can gain access to the bloodstream via three major routes: through the GI tract mucosal barrier, from an intravascular catheter, and from a localized source of fungal infection.^2–7 A prolonged stay in an intensive care unit is associated with an increased risk of fungal infection. Depending on the patient population and the rigor and thoroughness with which sites are cultured, as many as 50% to 80% of critically ill patients may experience Candida colonization at a single site.^2–7 The colonization index (the number of sites tested divided by the number of sites colonized) and the corrected colonization index (the number of sites tested divided by the number of sites showing heavy growth) have been proposed as criteria for selecting a high-risk patient population to be considered for possible early preemptive or therapeutic intervention.^3,^8,9 When test results from multiple sites show no evidence of fungal colonization, it is almost certain that no fungal infection is present.^3,^6,⁹

At a basic level, fungal invasion or infection might be defined as isolation of a fungal pathogen from a patient with signs and symptoms of an infection. As applied to fungal pathogens, however, this definition is an oversimplification. Clinicians relying on such a simplified definition would be prone to mistake some instances of fungal colonization for cases of true fungal infection. A degree of consensus has been reached regarding what constitutes a fungal infection in a neutropenic patient,¹⁰ but to date, no such consensus has been achieved as to what constitutes fungal infection in a nonimmunocompromised patient.^2,^8,^11–18

Given that it is not possible to establish a definitive diagnosis of fungal infection in all cases, it is often reasonable to express the diagnosis in terms of the degree of certainty about the presence of infection. Fungal infections can thus be classified as definite or probable.¹⁰ In brief, a definite fungal infection is present when deep tissue invasion is seen on a biopsy or culture from a sterile site in conjunction with clinical or radiologic abnormalities.¹⁰ Unfortunately, there is currently no agreement as to what signs and symptoms constitute a probable or possible fungal infection in a nonneutropenic patient.^16–18

A positive blood culture has traditionally been the gold standard for the diagnosis of candidemia and is now universally accepted as establishing the presence of a fungal infection.^16,17 Because blood-borne Candida can originate from a variety of different sites—including deep-seated infections (via hematogenous spread), catheter-related infections, and the gastrointestinal tract (via translocation), to name but a few—it is imperative that clinicians evaluating candidemic patients search carefully for all potential sources of infection. The search should include assessment of intravascular catheter sites, evaluation for peripheral septic thrombophlebitis, and examination of the abdomen as a possible site of infection. Because metastatic infection may develop from candidemia, new sources of fungal pathogens (e.g., endophthalmitis, infective endocardititis, hepatosplenic candidiasis, and arthritis) may arise over time.^16,17

A positive blood culture is an indication for initiation of antifungal therapy.^16,17 However, systemic Candida infection may be present even in the absence of a positive blood culture. Blood cultures have a 50% false-negative rate and may take as long as 4 days to yield results. Accordingly, clinicians should not assume that the absence of a positive blood culture or tissue biopsy means that a fungal infection is not present; nearly 44% of patients with a negative blood culture have evidence of disseminated fungal infection at autopsy.¹⁹ Increasingly, candidal invasion is being observed on autopsy specimens in cases where the diagnosis had not been clinically suspected.²⁰

Patients in whom several noncontiguous organs are infected by Candida species are considered to have a disseminated infection acquired through hematogenous spread. To establish the diagnosis, the organism must be identified by means of histologic analysis, culture of tissue samples obtained from at least one internal organ, or both, and there should be radiographic, pathologic, or culture-derived evidence of infection in at least one other organ. The presence of candidemia in association with Candida skin lesions or of endophthalmitis in a setting consistent with a diagnosis of Candida infection is also diagnostic of disseminated candidiasis.¹⁷ Hepatosplenic candidiasis is a more chronic form of disseminated Candida infection that typically develops in neutropenic cancer patients.¹⁷

Hospital-acquired, invasive Candida infections have become increasingly common causes of morbidity and mortality in hospitalized patients. Between 1979 and 2000, the incidence of Candida bloodstream infections increased by 207%.^1,^21,22 Data from the National Nosocomial Infections Surveillance System (NNIS) indicated that in the period from 1989 to 1998, C. albicans was the seventh most common cause of nosocomial infection in the ICU setting, accounting for 4.9% of bloodstream infections and 4.8% of surgical site infections.²³ Data from another national surveillance study, Surveillance and Control of Pathogens of Epidemiological Importance (SCOPE), indicated that in the period from 1995 to 2002, Candida species, as a group, were the fourth most common cause of nosocomial bloodstream infection (after coagulase-negative staphylococci, Staphylococcus aureus, and enterococci).²⁴ In ICU patients, Candida species were the third most common cause of nosocomial bloodstream infection. Between 1995 and 2002, the percentage of bloodstream isolates accounted for by Candida species rose from 8% to 12%.

Population-based studies have identified certain demographic groups that appear to be at special risk for invasive Candida infection.^1,²⁵ One such group consists of patients at the extremes of age, with neonates being one of the most rapidly growing at-risk groups. Race appears to play a role as well. In one study, the incidence of fungal infections in neonates younger than 30 days was reported to be 466/100,000 (960/100,000 in black neonates and 238/100,000 in white neonates).²⁵ Surgical patients are also at particular risk for fungal infection: more than 50% of all fungal infections occur in this patient population.²⁴ In the National Epidemiology of Mycosis Survey (NEMIS), 42 Candida bloodstream infections were identified in 4,276 patients admitted to six surgical ICUs between 1993 and 1995, for an incidence of 9.8 such infections per 1,000 admissions.²⁶Candida species caused 9.2% of all bloodstream infections diagnosed in the surgical ICUs, and more than half of the isolates were non-albicans species.

The morbidity and mortality associated with fungal infections are considerable.^27–30 Between 1980 and 1997, multiple-cause mortality from mycoses in the United States rose from 1,557 deaths to 6,534 (most of which were associated with Candida, Aspergillus, and Cryptococcus species).²⁷ Two case-control studies performed at a single institution 15 years apart revealed no improvement in the mortality attributable to candidemia: the attributable mortality was 38% in the earlier study (1983–1986)³¹ and 49% in the later one (1997–2001).³²

Moreover, fungal infections place a substantial economic burden on the health care system.^28–31,³³ In one study, costs were, on average, $41,000 (in 1993 dollars) higher for high-risk ICU patients than for low-risk ICU patients.³⁰ In a large European study, patients with Candida infection stayed longer both in the ICU (12.7 days) and in the hospital (15.5 days), and the direct costs associated with their care were higher by almost 16,000 euros.³⁰ In an earlier study from my institution (Johns Hopkins Hospital), the attributable increase in the cost of ICU care for surgical patients with fungal infections was $21,590 (in 1996 dollars).²⁹

The risk factors associated with the acquisition of fungal infections have been well defined [see Table 1].² ^11–14,^34,35 General factors contributing to both the risk of incurring a fungal infection and the outcome to be expected after such an infection include disruption of cutaneous or mucosal barriers, defects in the number and function of neutrophils or in cell-mediated immunity, metabolic derangements (as seen in preoperative and postoperative patients), and extreme youth or advanced age (see above). In an effort to create a rule for identifying a patient population with at least a 10% risk of fungal infection, a database of 2,890 patients who had stayed in an ICU for more than 4 days was subjected to retrospective review and statistical modeling.¹⁴ The best rule the authors were able to formulate included either the use of a systemic antibiotic or the presence of an intravenous central line, along with any two of the following: total parenteral nutrition, any dialysis, surgery, pancreatitis, or immunosuppressive agents. This rule captured 34% of patients with fungal infections. More important, it had a high negative predictive value and could be used to help exclude fungal infection as a likely possibility. Several other authors have attempted to predict a high-risk patient population, but to date, these models have not been used successfully in published papers or clinical trials and thus have not entered clinical practice.^8,¹⁵

Of surgical patients, those in the ICU, those who have sustained trauma or burns, and those who have received solid organ transplants (liver, pancreas, and small bowel transplant recipients, in particular) appear to be at higher risk for invasive fungal infection. The incidence of invasive Candida infection in liver transplant patients ranges from 1.3% to 15% for those receiving antifungal prophylaxis^36–38 and may be as high as 23% for those not receiving prophylaxis.³⁹ In pancreas transplant recipients, the incidence of invasive fungal infection is approximately 9%⁴⁰; in small bowel transplant recipients, it may be as high as 59%.⁴¹ Longer stays in the ICU, GI surgery, and the presence of invasive central venous lines appear to impart special risk to surgical patients.^3,^26,⁴² One of the most important steps in treating fungal infection is recognizing that a particular patient is at risk for this condition.^16–18 As noted [see Definition and Classification, Fungal Colonization versus Fungal Infection, above], fungal colonization is certainly an important risk factor for invasive fungal infection.^2–7

Whereas most fungal infections probably originate from endogenous sources, some undoubtedly originate from interactions with patients' external environments—in particular, from interactions with health care workers.^43–47 Cross-infection with Candida via hand transmission has been described in several studies.^43–47 In one study, 6% of nurses' hand cultures contained fungal pathogens, with which 2 of 57 patients with candidemia could be linked by means of DNA fingerprinting.⁴⁸ In another, nail contamination in the operating room was linked to an outbreak of C. tropicalis-infected sternal wounds.⁴⁷ Parenteral nutrition has also been identified as a potential source of contamination with Candida.⁴⁹

Patients with candidemia or disseminated candidiasis can present with a wide variety of clinical signs and symptoms.^11,^17,18,⁵⁰ Identification of high-risk patients is probably the single most important step in establishing the diagnosis of candidiasis.^8,^14,15,^17,18,^36,⁵⁰ Although they are not very common presenting signs or symptoms, candidal endophthalmitis, suppurative phlebitis, and candiduria in the absence of bladder instrumentation may each provide special clues suggesting an increased risk for fungal infection. A careful eye examination to identify the presence of candidal infection should be performed while the results of cultures of various sites are being awaited [see Investigative Studies, Culture, below], and a repeat examination should be performed after therapy for proven candidemia. Candidal endophthalmitis may remain asymptomatic until late in the course of infection.^17,18,^50,51

The presence of peripheral suppurative phlebitis that fails to yield bacteria or does not respond to antibacterial agents may be an early clue to the presence of hematogenous candidiasis. Gentle squeezing of the venous catheter exit site may express pus that yields Candida species on a smear or culture. Surgical excision of the infected vein usually reveals Candida infection in its lumen.^17,18,^50,⁵²

The presence of high-grade candiduria in surgical patients who have not undergone instrumentation of the renal pelvis or the bladder suggests hematogenous candidiasis and should prompt a workup for this infection. In this setting, candiduria may result from seeding or filtering through the kidney.^17,18,^50,^53–58

Fever is sometimes the only sign of infection; however, it may be absent in infected patients who are receiving corticosteroids. Because fungal pathogens interact with the Toll-like receptors (TLRs), just as bacterial pathogens do, patients with fungal infections may present with the systemic inflammatory response syndrome (SIRS) or septic shock. Accordingly, the diagnosis should be seriously considered in high-risk patients who are persistently febrile. Because of the high mortality associated with fungal infection, empirical antifungal therapy is recommended for early treatment of a clinical occult infection or for prevention of a new infection.

The workup of a surgical patient with suspected hematogenous candidiasis begins with a complete set of cultures of sputum, oropharynx, stool, urine, all drain sites, and blood [see Table 2]. Candidiasis rarely develops in patients whose cultures show no evidence of Candida at any site.^6,7,⁴³ Obtaining more than six sets of blood cultures has little value, and there is no evidence to support obtaining arterial cultures. Once a blood culture has turned positive for Candida, the species must be rapidly identified so that the correct antifungal agent can be selected for treatment. A rapid and inexpensive test is the germ tube test, which can distinguish C. albicans (positive result) from other Candida species. More than 90% of C. albicans isolates produce germ tubes when incubated in serum for 2 to 3 hours at 37° C. Several commercial products are now available to aid in the rapid identification of Candida species.^59,60

Positive cultures from nonsterile sites (sputum, urine, and wound drainage) must be interpreted with caution because of the frequent occurrence of Candida as a normal commensal of humans. Such cultures are useful mainly as an indication of colonization and, consequently, of the risk of infection in the appropriate setting.

Because of the emergence of non-albicans species as significant pathogens, it is vital that surgeons be aware of the variety of Candida species that may be encountered and the relevant drug sensitivities.^80–83 Most C. albicans isolates are sensitive to all of the currently available agents, but some low-level resistance has been reported, especially with previous long-term exposure to azoles at low dosages.^80–83C. glabrata may be resistant to fluconazole or require some modification of the dosage; C. krusei is resistant to fluconazole; and C. lusitaniae may be resistant to amphotericin B.^80–83 The selection of the initial antifungal agent to be used is based on estimation of the likelihood of infection, assessment of the probable sensitivity or resistance of the fungal species, consideration of specific patient risk factors (e.g., renal or hepatic insufficiency), and determination of whether the patient should receive the agent orally or intravenously.^68–79 In addition, selection of the best antifungal agent for a specific at-risk or infected patient is facilitated and enhanced by familiarity with the typical pathogens in the local community.

Several oral and intravenous azoles are available for clinical use.^17,18,^69,^72,^74–77 Clotrimazole and miconazole are not suitable for systemic therapy. Ketaconazole has very poor and erratic GI absorption and thus is not used for systemic therapy. Itraconazole also has very erratic GI absorption and therefore should not be used for systemic therapy either.^17,18 A single trial conducted in a pediatric population found that itraconazole at a dosage of 10 mg/kg/day orally was clinically similar to fluconazole at a dosage of 200 mg/day orally for treatment of candidemia.⁸⁴ The I.V. agent fluconazole is a well-tolerated triazole that shows good activity in candidemic patients. Individual clinical trials involving nonneutropenic patients with candidemia, as well as a meta-analysis of these trials, indicate that amphotericin deoxycholate and fluconazole yield similar outcomes in terms of overall mortality, attributable mortality, and clinical and microbiologic response but differ in terms of toxicity, with fluconazole appearing to be less toxic.^85–88 Thus, it is reasonable to use fluconazole to treat patients with candidemia.

The 2004 treatment guidelines published by the Infectious Disease Society of America suggest that physicians may choose among three alternatives for primary treatment of candidemia in nonneutropenic adult patients: (1) amphotericin B deoxycholate (0.6 to 1.0 mg/kg/day, or one of the lipid formulations); (2) fluconazole (400 to 800 mg orally or I.V.); and (3) caspofungin (loading dose of 70 mg I.V., followed by 50 mg/day I.V. or, if the patient has severe obstructive hepatic disease, 35 mg/day I.V.).¹⁷ If the species is known or believed to be C. albicans, fluconazole (loading dose of 12 mg/kg I.V., followed by I.V. infusion of 6 mg/kg/day) is appropriate. Fluconazole has excellent oral bioavailability and frequently can be delivered orally after the first few days. Dose adjustments must be considered if renal insufficiency is present.^17,18,^69,^72,^74–77 If the patient has received long-term azole therapy, is known to be colonized with C. glabrata or C. krusei, or both, then one of the echinocandins—either caspofungin (loading dose of 70 mg I.V., followed by 50 mg/day I.V.) or micafungin (100 mg/day I.V.)—should be administered.^68,^71,^73,^78,⁸⁹ If the species then proves sensitive to fluconazole, this agent may then be given orally to continue therapy. Although clinical trials have shown voriconazole (6 mg/kg I.V. every 12 hours for the first 24 hours, then 3 mg/kg I.V. every 12 hours, p.o. switch 200 mg every 12 hours) to be an acceptable agent for treating candidemia,⁹⁰ there are many drug interactions and side effects of which the physician must be aware when using this drug. In particular, patients who have undergone transplantation and those with HIV infections are very likely to be taking agents that interact with voriconazole.

The current role of amphotericin B in the treatment of candidemia is limited by its toxicity [see Systemic Antifungal Agents, Amphotericin B, below]. The lipid-associated formulations of amphotericin B are less nephrotoxic than the parent compound.⁶¹ Three such formulations are available: (1) amphotericin B lipid complex (Abelcet), (2) amphotericin B colloidal dispersion (Amphocil, Amphotec), and (3) liposomal amphotericin B (AmBisome). A prospective, randomized trial found Abelcet to be as efficacious as conventional amphotericin B for treating hematogenous candidiasis.⁶² If, however, the patient is known to have a Candida infection, the infection can and should be treated with one of the agents previously mentioned. In a 2005 cost-effectiveness analysis, both amphotericin B deoxycholate and its lipid products provided fewer discounted lives saved than either fluconazole or caspofungin did (according to the data available at the time of publication).⁸⁹

At present, there are no evidence-based guidelines concerning the optimal duration of therapy.^17,18 This decision should be based in part on the severity of the patient's illness, as well as on his or her immune constitution. In general, therapy should be continued for at least 14 days after the last positive cultures and the resolution of all signs and symptoms of infection in the most seriously ill patients. Shorter courses of therapy may be appropriate in carefully selected patient populations without signs of hemodynamic instability or evidence of distant metastatic infection.

The fundamental controversy about whether central venous access lines should be removed when identified in patients with candidemia stems from the multiple routes by which infection occurs in different patient populations.^91–94 When the GI tract is the source of candidemia (as in cancer patients with mucositis),⁹⁵ catheter removal is unlikely to yield any significant benefit. On the other hand, when an I.V. catheter is a source of ongoing candidemia, there is some evidence to suggest that removal of the catheter may shorten the time to blood clearance and perhaps even improve overall survival. In a study examining catheter exchange in patients without cancer, replacement of all vascular catheters shortened the duration of candidemia from 5.6 days to 2.6 days.⁹¹ The current state of knowledge regarding the importance of biofilm, the penetration of antifungal agents, and the attachment of new pathogens tends to favor removal of catheters to the extent possible after candidemia. Given the available data, physicians should consider removing all intravascular catheters from inpatients who have candidemia with persistent fever, persistent or high-grade fungemia, or C. parapsilosis infection (which is more likely to be catheter related than infection with another Candida species would be).

The use of antifungal prophylaxis remains a controversial topic,^3,^99–108 though four recent meta-analyses^109–112 and a Cochrane review¹¹¹ have all concluded that prophylactic administration of antifungal agents to high-risk critically ill or surgical patients reduces the incidence of invasive fungal infection but has no proven effect on mortality. These analyses illustrate the problems inherent in attempting to combine data from small trials that employed different agents (ketoconazole and fluconazole) at different dosages in different at-risk patient populations. Perhaps the most difficult issue in determining whether antifungal prophylaxis should be employed involves the lack of a consensus on how fungal infection should be defined. Without a well-understood end point, it is impossible to decide whether a therapy should be widely applied.

To date, there have been eight major studies of antifungal prophylaxis that are applicable to the critically ill surgical patient population.^3,^102–108 The three largest of these are of particular relevance. In one study, a large number of liver transplant recipients were randomly assigned to receive either placebo or fluconazole for a period of 10 weeks (during which administration switched from I.V. to oral).¹⁰⁵ Definitions of fungal infections included both deep and superficial infections. The authors documented a significant decrease in fungal colonization and fungal infection (both superficial and deep) in the fluconazole group but reported no significant difference in mortality between groups. Although hepatotoxicity was absent in the fluconazole group, neurologic toxicity was higher than in the placebo group, presumably as a consequence of increases in cyclosporine levels, which must be adjusted in patients receiving prophylactic fluconazole.

In a large study that enrolled 220 patients in the medical or surgical ICU on or after their third day in the unit, prophylaxis with fluconazole, 100 mg I.V., was compared with placebo.¹⁰⁸ The patients enrolled in this trial were critically ill, were receiving respiratory support from a mechanical ventilator, and were undergoing selective decontamination of the GI tract. Candidemia was virtually eliminated in the fluconazole group. In addition, the incidence of invasive candidal infection was lower in this group (3.9% versus 8.9%), and fungal colonization was both less frequent and less intense.

Finally, in a single-institution, randomized, double-blind, placebo-controlled trial that compared enteral fluconazole (800 mg loading dose, followed by 400 mg/day) with placebo for prevention of fungal infections in high-risk critically ill surgical patients, the authors reported that fluconazole prophylaxis yielded a two- to threefold reduction in fungal infection but had no effect on mortality.³ Using a strict case definition that did not include fungal colonization, the intent-to-treat analysis demonstrated 20 patients with fungal infections (15.3%) in the placebo group and 11 such patients (8.5%), including four with infections that were unknown at enrollment, in the fluconazole group. The number needed to treat (i.e., the number of patients who would have to receive antifungal prophylaxis for a single fungal infection to be prevented) was 14.5, a low number that suggests a highly significant effect.

The data do not yet support widespread or routine use of antifungal prophylaxis. Accordingly, it is worthwhile to formulate some general conclusions regarding which patients should receive antifungal prophylaxis and which agent should be used. It is reasonable to consider antifungal prophylaxis in patients similar to those in the aforementioned studies (see above), including organ transplant patients^3,^17,^109–112 and patients enrolled in referenced clinical trials. In most of these patient populations, fluconazole is the most appropriate agent for prevention of Candida infection: it has few side effects, it can be administered orally, and out of the agents that have been well studied, it is the most reasonably priced. In view of the increasing reports of resistance, however, the use of fluconazole should not be extended to additional patient populations, especially those who are less ill and certainly those who are at less risk. The key question in making the decision for or against antifungal prophylaxis should be whether the benefits achievable in the patient population being considered for prophylaxis are significant enough to outweigh the cost and the risk of eventual (and even expedited) resistance developing with excessive use.

Fluconazole is currently the drug of choice because of its proven efficacy and its higher concentration (20% to 70% of the corresponding plasma level) in ocular tissue, including the vitreous body.^51,¹¹⁴ Accordingly, I recommend 800 mg/day of fluconazole until a major response is observed, at which time it may be possible to reduce the dose to 400 mg/day. Many other Candida and non-Candida fungal species may also cause endophthalmitis, and in such cases, fluconazole may not provide effective coverage. If the infecting organism (especially C. krusei) is potentially resistant to fluconazole, the recommended therapy is amphotericin B (0.7 to 1.0 mg/kg/day I.V.), preferably in conjunction with flucytosine.¹¹⁵ Intravitreal injection of amphotericin B is recommended for vitreal infections. Both voriconazole and caspofungin have had limited success in patients with more resistant infections.⁵¹

The optimal duration of therapy for endogenous endophthalmitis is unknown. Ophthalmologic consultation is critical in establishing the diagnosis, assessing the patient's response to therapy, detecting complications, and determining whether early vitrectomy is indicated to prevent loss of sight.¹¹⁶

A rare but serious consequence of hematogenous candidemia is suppurative thrombophlebitis, which results from infection of a vessel traumatized by prolonged catheterization. Endothelial disruption exposes the basement membrane and leads to thrombus formation and propagation. Suppurative thrombophlebitis is particularly serious because intravascular infection results in a persistent high-density fungemia. Management of this disease consists of high-dose antifungal therapy, removal of the central venous catheter, and excision of the infected vein (when possible).^52,117 Typically, blood cultures remain positive for several days; sometimes, they remain positive for as long as 3 to 4 weeks despite appropriate antifungal therapy, if the infected vein is not excised.

Fungal infections account for between 1% and 10% of all pathogens isolated from patients with infective endocarditis.¹⁷ In patients with prosthetic valves, the rate at which fungi are isolated may be as high as 10%. The overall outcome of candidal infective endocarditis is grim, carrying a reported mortality of up to 80%, especially when complicated by congestive heart failure and persistent fungemia.¹¹⁸ Embolic phenomena are more commonly associated with fungal endocarditis than with the nonfungal variety.

Candidal endocarditis is very difficult to treat. A 2005 review found that patients with candidal endocarditis who underwent adjunctive surgery had a lower reported mortality.¹¹⁸ Thus, surgical replacement or repair of valves is required to prevent death from embolization or cardiac failure in patients with complicated prosthetic valve endocarditis. Amphotericin B, with or without flucytosine, has been the therapy of choice; however, neither the optimal dosage nor the optimal duration of therapy has been determined. There is some anecdotal information suggesting that fluconazole, with or without caspofungin, may be successfully employed in this setting, primarily as long-term suppressive therapy after the initial administration of amphotericin B.¹¹⁹ Because of the high risk of late relapse, long-term maintenance therapy with oral fluconazole should be considered in patients unwilling or unable to undergo surgical treatment.

Candidal pericarditis is a very rare complication of hematogenous candidiasis. The surgical patients at risk for purulent pericarditis caused by Candida are those who have undergone a cardiac operation, those who have a malignancy and whose host defenses are impaired, and those who have a debilitating chronic disease.¹²⁰ Patients should be treated with surgical drainage or debridement and with high-dose amphotericin B or fluconazole.

Candidal joint infections can develop via hematogenous spread as a consequence of inadvertent direct inoculation during joint procedures or intra-articular injection of corticosteroids. Such infections typically involve a single joint (most frequently the knee or the hip) and tend to occur in patients with rheumatoid arthritis or prosthetic joint devices.¹²¹ Local symptoms (i.e., pain on weight bearing or on full extension) may be present. Diagnosis is best achieved by visualizing the organisms or growing them from the joint fluid. Early diagnosis and systemic antifungal therapy are vital for preventing destruction of the cartilage or loosening of the prosthesis.

Successful treatment of fungal arthritis with fluconazole has been reported in several cases.¹²² I recommend giving fluconazole at a dosage of 400 to 800 mg/day for 6 to 12 months; others suggest giving amphotericin B first for 1 to 2 weeks, then switching to fluconazole.¹⁷ Fluconazole can be used for short-term therapy, either alone or in combination with surgery, as well as for long-term suppressive therapy in patients who are at risk for recurrence or who cannot undergo surgical debridement. In general, surgical resection of the involved joint is required.

Except for sternal infections complicating median sternotomy, most cases of candidal osteomyelitis develop through hematogenous spread. Vertebral body involvement is common. Back pain and fever may be followed by radiculopathy. Surgical drainage of all purulent collections is essential for a good response; however, surgical debridement of bony lesions may not be needed. Although amphotericin B has been the standard drug of choice for candidal osteomyelitis, fluconazole may be considered as an alternative. Amphotericin B should not be used for mediastinal irrigation after drainage and debridement of a mediastinal infection, because there is a high probability of chemical mediastinitis.

Candidal meningitis may follow hematogenous spread, or it may be a complication of neurosurgery or the implantation of ventriculoperitoneal shunts. In neonates, Candida infection may be present without previous surgery and without previously documented fungemia, but often, an intravascular catheter is present.¹²³ The infection is insidious and sometimes goes undiagnosed. Most patients with candidal meningitis have recently received antibacterial agents, and half have previously had bacterial meningitis. The overall mortality is around 10%.

The standard therapeutic regimen for candidal meningitis consists of amphotericin B with flucytosine. As an alternative, a combination of high-dose fluconazole (800 mg/day) and flucytosine (50 mg/kg/day) may be considered; this is a particularly attractive approach because of the high CSF concentrations achieved with both agents. Removal of the infected shunts, as well as the intravascular catheter, is recommended when possible.^17,¹²³ The duration of treatment should be based on the patient's clinical response and the culture results. In general, prolonged treatment (more than 4 weeks beyond the resolution of symptoms) is suggested.

True candidal pneumonia is rare, but it can occur through hematogenous dissemination into the lung as one of many sites of infection.^17,^124,125 Oropharyngeal aspiration of oral and upper tracheal organisms does not typically cause pneumonia. The value of deep tracheal suction or even bronchoalveolar lavage in defining this disease should be questioned. Histopathologic confirmation is recommended.¹⁷ Laryngeal infection should be considered; some case reports suggest that it can be rapidly progressive.

Nystatin should be given as a 10 to 30 ml suspension five times daily, and the patient should be instructed to swish it around the mouth before swallowing; alternatively, the patient may take one or two troches five times daily. Clotrimazole is given five times daily in the form of a 10 mg troche that should be held in the mouth until dissolved. In surgical patients at risk for hematogenous infection, systemic therapy with ketoconazole (200 to 400 mg once daily), itraconazole (200 mg/day), or fluconazole (100 mg once daily) is preferred.¹⁷ Antifungal therapy should be administered for 7 to 14 days. Ketoconazole and itraconazole are slightly less effective than fluconazole because of variable GI absorption. Repeated episodes of oral candidiasis are common.

In a minority of patients, the pathology of candidal infection of the lower GI tract may change from diarrhea without demonstrable tissue invasion to direct penetration into the submucosa, which eventually leads to pseudomembranous enterocolitis. Direct vascular invasion through the bowel wall may occur in immunosuppressed patients; these patients may exhibit extensive involvement of the GI tract from mouth to anus. Nonneutropenic surgical patients may exhibit more localized GI involvement.

The preferred therapy for esophageal candidiasis consists of fluconazole (100 mg/day orally) or itraconazole (200 mg/day) for 14 to 21 days.¹⁷ In patients who remain symptomatic after 7 days of therapy, endoscopy is indicated to rule out other causes of esophagitis. If endoscopy proves that esophagitis is being caused by candidal infection, low-dose amphotericin B (0.4 mg/kg/day) may be administered. Alternatively, caspofungin or voriconazole may be given,¹⁷ though it should be kept in mind that voriconazole causes more adverse events without being more efficacious. Therapy for esophageal candidiasis should be continued for at least 4 days after symptoms resolve; immunosuppressed patients generally require more extended therapy to prevent relapse.

Because stool cultures do not differentiate between colonization and infection, candidiasis in the lower GI tract is usually a postmortem diagnosis; hence, there are no reliable criteria governing when and how to treat this condition. It has been reported, however, that patients who have diarrhea that can only have been caused by heavy colonization with Candida species may respond dramatically to 2 to 4 days of nystatin therapy.

Perhaps the most controversial issue in the management of Candida infectious syndromes in surgical patients is the question of whether specific systemic therapy is required to eradicate the infection within intra-abdominal abscesses, peritoneal fluid, or fistula drainage.¹⁷Candida is frequently cultured from intra-abdominal infectious foci but should be considered a serious threat only in high-risk patients.¹²⁶ Four risk factors for intra-abdominal candidiasis have been identified: (1) gastrointestinal perforations, especially in the upper GI tract, (2) anastomotic leakage, (3) surgery for acute pancreatitis, and (4) splenectomy.^107,¹²⁶ In patients with a localized fungal infection necessitating drainage, there is no replacement for adequate drainage and debridement. Such infections are often located in the abdomen, the liver, the biliary tree, the pancreas, or the peritoneal cavity. Percutaneous drainage can play an important role in the management of these difficult problems; however, one retrospective review of percutaneous drainage of fungal collections in the abdomen or thorax suggested that clinical failures occur more often when complex fluid collections are visible radiographically, when patients have a history of malignancy, and when patients are critically ill.¹²⁷

Patients with peritonitis and intra-abdominal abscesses should receive systemic antifungal therapy, usually in combination with antibacterial therapy (given that these infections are almost always polymicrobial in origin). The risk of dissemination is increased both by the recurrence of intra-abdominal infection and by the presence of extensive areas of communication between the abdominal cavity and the external environment via either fistulas or drain tracts. Because fluconazole is very safe and is capable of reaching high concentrations in peritoneal fluid, it is useful in the management of candidal peritonitis and should be considered the agent of choice in cases where the pathogens are susceptible to it.¹⁰⁷ Fluconazole should be given at a dosage similar to that employed in treating systemic disease (loading dose of 12 mg/kg I.V., followed by 6 mg/kg/day I.V.), with a 50% dosage reduction in cases of renal impairment. When pathogens are resistant to fluconazole, caspofungin and voriconazole at systemic dosages are acceptable alternatives.¹⁷

On rare occasions, candidal peritonitis occurs after long-term ambulatory peritoneal dialysis. This infection is not clinically distinguishable from bacterial peritonitis. In such cases, the infection tends to remain localized and to manifest with low-grade fever and abdominal pain and tenderness. The peritoneal dialysate is usually cloudy and contains more than 100 neutrophils/mm³. Therapy consists of systemic antifungal therapy and removal of the peritoneal catheter. The abdominal pain caused by the addition of amphotericin B to the dialysate has raised concern that chemical peritonitis might give rise to adhesions and thus impair the efficacy of dialysis. Accordingly, this agent should not be used for peritoneal irrigation. Immediate removal of the peritoneal catheter has been recommended. Fluconazole should be considered as the agent of choice and should be given for 1 to 14 days. A new catheter should not be placed for 4 to 6 weeks. In centers with a high rate of fungal peritonitis secondary to peritoneal dialysis, nystatin prophylaxis may be considered. A prospective, randomized study of patients on continuous ambulatory dialysis found that Candida peritonitis was successfully prevented with oral nystatin (tablets containing 500,000 units given four times a day).¹²⁸

Occasionally, Candida species may cause cholangitis, biliary tract disease, pancreatic abscess, or liver abscess. This problem is increasingly found in cancer patients with percutaneously placed drainage catheters. Such patients must be given systemic therapy if there is any clinical evidence of infection (e.g., candidemia), and the drainage catheter must be changed. Diverticulitis complicated by candidal pylephlebitis has also been reported.

Diagnosis and treatment of candidal wound infections are problematic. Recovery of Candida species from drains and wounds does not necessarily mean that this organism is causing tissue infection; however, isolation of Candida or demonstration of invasive forms on biopsies at reoperation would suggest that invasive infection is present. Antifungal therapy should be administered to patients with demonstrated invasive infection and possibly to those in whom antibacterial therapy has failed and Candida has been isolated. If a deep sternal wound culture is obtained and is positive for Candida after coronary bypass surgery, antifungal agents should be given to prevent the establishment of candidal osteomyelitis of the sternum.

Besides simple colonization, the spectrum of candidal urinary tract infection includes cystitis, pyelitis (i.e., infection of the renal pelvis), fungus ball of the ureter, and renal abscesses. The diagnosis of cystitis is based on the presence of symptoms of cystitis, diffuse erythema or fungal plaques on cystoscopy, candiduria, and pyuria. Candida cystitis warrants therapy. If a triple-lumen catheter is in place, bladder irrigation with amphotericin B (50 mg/L/day for 2 days) may be tried. Fluconazole (200 mg once daily) is a more attractive approach because of the convenience, the lower cost, and the very high drug concentrations achieved in the urine.⁵⁴ Flucytosine is excreted in the urine in high concentrations and may be particularly useful against C. glabrata infection.

Management of pyelitis depends on whether the renal papillae are invaded or the ureter is obstructed with a fungus ball. Patients with no papillitis and an open ureter usually respond to irrigation with amphotericin B through a ureteral or percutaneous catheter. If papillitis is present, systemic antifungal therapy with fluconazole (400 mg/day) is recommended. If fungus balls are present, surgical removal should be considered in addition to treatment with antifungal agents.⁵⁴

Hematogenous infections of the kidneys leading to multiple renal abscesses are treated as instances of hematogenous candidiasis. However, because both fluconazole and flucytosine are well tolerated, have good tissue diffusion, and are excreted in the urine in high concentrations, it is reasonable to assume that the combination of these two agents represents the therapy of choice for such infections.

Acute invasive pulmonary aspergillosis is the most common form of Aspergillus infection in immunocompromised surgical patients.⁸² The organisms tend to invade blood vessels and cause thrombosis and infarction of the surrounding tissues. The infection may manifest itself in the form of acute vascular events such as pulmonary embolisms or, more rarely, myocardial infarction, cerebral hemorrhage, or Budd-Chiari syndrome. Pulmonary Aspergillus infections include necrotizing bronchopneumonia and hemorrhagic pulmonary infarction, each accounting for about one third of these infections. Pulmonary aspergillosis may extend to contiguous organs or be disseminated. The rhinocerebral form of aspergillosis occurs less often than the pulmonary form. It originates in the sinuses and progresses through soft tissues, cartilage, and bone, causing lesions in the palate and the nose. Occasionally, the infection progresses through the base of the skull and involves the brain.

Diagnosis of Aspergillus infection is difficult. Computed tomography can be diagnostic¹³¹ and often allows early recognition of aspergillosis. In one study that used CT scans as a means of investigation, the mean time to diagnosis of invasive pulmonary aspergillosis fell from 7 days to less than 2 days, and this reduction was associated with a decrease in mortality. CT may also be combined with detection of Aspergillus antigen (galactomannan), an approach that shows promise for early diagnosis of invasive pulmonary aspergillosis.¹³² Often, however, diagnosis of Aspergillus infection depends on identifying the organism in culture or his to pathologic specimens.⁸² Recovery of Aspergillus species from the respiratory tract culture of a surgical patient should be considered to represent contamination or colonization unless the patient is symptomatic and severely immunosuppressed.

All of the antifungal agents currently used for primary treatment of Aspergillus are associated with relatively high incidences of treatment failures and fairly poor outcomes. At present, voriconazole (at the dosage previously described [see Management of Candidemia and Acute Disseminated Candidiasis, above]) is the treatment of choice, with lipid amphotericin B formulations as secondary agents. In some institutions, combination therapy with an echinocandin is occasionally employed as an alternative.

Some of the most rapidly advancing forms of angioinvasive fungal infection are caused by the Zygomycetes Mucor and Rhizopus.¹³³ The reservoir, the mode of transmission, the pathogenesis, and the clinical presentations are similar to those of Aspergillus species. Zygomycosis is being seen with increasing frequency in stem cell transplant recipients who have been receiving voriconazole prophylaxis. In surgical patients, the major risk factors include diabetic ketoacidosis, immunosuppression after cytotoxic chemotherapy, adrenal corticosteroid therapy, organ transplantation, skin damage (e.g., from adhesive tape, an arm board, or severe burns), iron overload, and a prolonged postoperative stay.¹³³ Zygomycotic infections may be rhinocerebral, paranasal, pulmonary, GI, cutaneous, or (very rarely) disseminated. Any patient with a necrotic lesion on the skin, soft, hard palate or nose should be considered to have an angioinvasive mold infection until emergency biopsy proves otherwise.

Appropriate management of zygomycosis consists of extensive surgical debridement of infected areas, rapid correction of the underlying disease, and administration of an appropriate antifungal agent. One of the lipid amphotericin products should be the first-line therapeutic agent, with high-dose amphotericin B deoxycholate a second choice.

Many other mold species are also emerging as significant pathogens. In an at-risk patient, recovery of any fungus from any site should prompt an evaluation by an infectious disease specialist to determine the clinical significance of the isolate.

Amphotericin B is lipophilic and binds avidly to ergosterol, the principal sterol in the fungal cytoplasmic membrane. This binding disrupts the integrity of the membrane, resulting in leakage of intracellular contents and cell death. The clinical effectiveness of amphotericin B against fungi is believed to be attributable to the drug's stronger affinity for ergosterol (found in fungal cell membranes) than for cholesterol (the principal sterol found in mammalian cell membranes). Another proposed mechanism of action is oxidation-dependent amphotericin B-induced stimulation of macrophages.⁹⁹ Most of the fungal species that cause human infections are susceptible to amphotericin B.

Because amphotericin B is lipophilic, it must be complexed with a bile salt (deoxycholate) to be able to stay in solution. Unfortunately, about 20% of patients receiving amphotericin B deoxycholate experience an acute infusion-related reaction consisting of fever, hypotension, and tachycardia.¹³⁵ Premedication with acetaminophen may blunt this response; if it does not, premedication with hydrocortisone (25 to 50 mg I.V.) or meperidine (25 to 50 mg I.V.) is recommended. Hypotension, hypertension, hypothermia, and bradycardia have also been reported as infusion-related toxic effects of amphotericin B deoxycholate. Ventricular arrhythmias have been associated with rapid infusion of amphotericin B deoxycholate and with administration of this agent to patients with severe hypokalemia or renal failure. Through inhibition of erythropoietic production secondary to nephrotoxicity, amphotericin B suppresses the production of red blood cells, causing a normocytic, normochromic anemia. As a result of renal tubular loss, acidosis, hypokalemia, and hypomagnesia are common. The need to replace lost electrolytes should be expected. Patients should be kept well hydrated.

Common practice is to give a 1 mg test dose and observe the patient for 1 hour in the hope of identifying patients at risk for severe acute reactions. The full dose of the drug (0.6 to 1 mg/kg/day) is then infused over a period of 4 to 6 hours, though there is evidence to suggest that much shorter infusion times (e.g., 1 hour in patients with adequate cardiopulmonary and renal function) may be acceptable. The total dose depends on the extent of the infection and the patient's condition. Patients must be monitored carefully during the first day of therapy. The infusion should be discontinued if the patient becomes hemodynamically unstable.

If acute reactions limit the amount of drug that can be infused, patients are premedicated with hydrocortisone (25 to 50 mg I.V.), either alone or in combination with meperidine (25 to 50 mg I.V.), 30 minutes before amphotericin B infusion is begun.

Renal toxicity and hypokalemia are the primary toxicities of amphotericin B [see Table 3]. Amphotericin B-induced nephrotoxicity may be glomerular (characterized by a decrease in glomerular filtration rate and renal blood flow) or tubular (with urinary casts, hypokalemia, hypomagnesemia, renal tubular acidosis, and nephrocalcinosis).¹⁰⁰ All of these abnormalities occur to varying degrees in almost all patients receiving the drug. In most patients, renal dysfunction gradually resolves after discontinuance of therapy. Amphotericin B nephrotoxicity may be minimized by refraining from using it with other agents that exert synergistic nephrotoxic effects¹³⁵ (e.g., aminoglycosides, vancomycin, cisplatin, and cyclosporine) and by providing sodium suplementation. Sodium supplementation consists of I.V. infusion of 500 ml of 0.9% saline solution 30 minutes before the administration of amphotericin B, followed by a second infusion of the same amount of saline after the amphotericin B infusion is completed.

To minimize synergistic nephrotoxicity when amphotericin B is used in conjunction with other nephrotoxic agents, use of one of the less nephrotoxic lipid formulations of amphotericin B—such as amphotericin B lipid complex (Abelcet), amphotericin B colloidal dispersion (Amphocil, Amphotec), or liposomal amphotericin B (AmBisome)—may be indicated.¹³⁶ These preparations differ from each other with respect to the amount of amphotericin B present and the type of lipid used, as well as with respect to their physical forms, pharmacokinetics, and toxicities. Of the three, liposomal amphotericin B (AmBisome) is the one that is best tolerated. It should be noted that all of these formulations do still have some negative effect on renal function.

Fluconazole is available in both oral forms (pill and suspension) and an I.V. form. In any form, it is rapidly and almost completely absorbed from the GI tract: serum concentrations achieved after oral administration are almost identical to those achieved after I.V. infusion. This high degree of GI absorption, which is not affected by gastric acidity or the presence of food, is a major advantage that fluconazole has over ketoconazole. An initial loading dose that is twice the usual daily dose is recommended. Fluconazole is distributed evenly in body tissues, penetrates into the vitreous humor and the aqueous humor of the eye, and crosses the blood-brain barrier. It is excreted largely unchanged in the urine and undergoes only minimal metabolism in the liver. Consequently, dosage schedules must be adjusted in patients with renal impairment. Hemodialysis substantially reduces serum concentrations of fluconazole, and peritoneal dialysis also appears to remove the drug. A standard dose should be given after each course of dialysis.

The toxic effects of fluconazole are similar to those of other azoles, including nausea and vomiting (in about 2% of patients), headache, fatigue, abdominal pain, and diarrhea^68–78; exfoliative dermatitis also occurs, but very rarely. Transient abnormalities of liver function are observed in approximately 3% of patients who receive fluconazole. Fatal hepatic necrosis was reported to have developed in two patients who were receiving fluconazole, but it was unclear whether the drug played a causal role. Fluconazole does not induce adrenal suppression. On occasion, fluconazole administration may lead to prolongation of the QT interval.

Because fluconazole interacts with warfarin, phenytoin, and cyclosporine when given in a daily dose of 200 mg or higher, serum concentrations of these agents should be monitored when they are used in conjunction with fluconazole.

Itraconazole is generally well tolerated in dosages as high as 400 mg/day. The most common side effects are nausea, vomiting, diarrhea, and abdominal discomfort. Headaches, rash, pruritus, and dizziness occasionally occur. At higher doses, a mineralocorticoid excess syndrome (characterized by hypokalemia, hypertension, and edema) has been described, and hypokalemia develops in as many as 6% of patients who take 400 mg/day for several months.¹⁰⁴

The Federal Drug Administration (FDA) has issued a black box warning against the use of itraconazole in patients with congestive heart failure and those at risk for drug-drug interactions. Because itraconazole may be teratogenic, it should be avoided in pregnant patients as well. Itraconazole is currently approved only for treating blastomycosis, histoplasmosis, and aspergillosis, but it is also effective for treating candidiasis and cryptococcosis.

Like the other azoles, posaconazole is well tolerated even with long-term administration.¹³⁷ The most common side effects are fever, nausea, vomiting, diarrhea, and headaches. Rash, thrombocytopenia, and abdominal pain occasionally occur. As a consequence of hepatic metabolism, posaconazole has been associated with elevated liver enzyme levels, hyperbilirubinemia, and hepatocellular damage. Liver function test results should be monitored at baseline and throughout the course of posaconazole therapy. Prolongation of the QT interval and adrenal insufficiency may also be seen with administration of this agent.

Because posaconazole is an inhibitor of cytochrome P-450 34A, it may be expected to increase the plasma concentrations of drugs metabolized via this pathway. Various drug interactions may occur, and it frequently proves necessary to adjust the dosages of other agents being given simultaneously with posaconazole. The adult dosage for posaconazole is 200 mg three times daily, taken with a full meal or a liquid nutritional supplement.¹³⁸ The use of this agent in pregnant patients is not recommended.

The echinocandins, as a class, are active against Candida species, including C. krusei and C. glabrata.^68–70,^73,⁷⁸ In addition, they are active against Aspergillus species, including A. fumigatus, A. flavus, and A. terres; in this setting, they show some synergism with itraconazole and voriconazole. None of the echinocandins are active against Cryptococcus, and none should be used for this indication. The emerging fungi Fusarium, Rhizopus, Mucor, Scedosporium, and Pseudallescheria boydii are all resistant to each of the echinocandins.

All of the echinocandins are available solely as I.V. formulations. Caspofungin is given in a loading dose of 70 mg I.V., followed by infusion of 50 mg/day.⁷⁰ Anidulafungin is given in a 200 mg loading dose on day 1, followed by infusion of 100 mg/day.⁷⁹ Micafungin does not require a loading dose and is given at a dosage of 100 mg/day.⁷¹ Caspofungin is slowly metabolized in the liver through nonenzymatic peptide hydrolysis and N-acetylation to yield two inactive metabolites. Micafungin is metabolized into three metabolites⁷¹; cytochrome P-450 3A plays a minor role in this process. Anidulafungin undergoes spontaneous degradation into an inactive open ring peptide; no hepatic metabolism has been observed.

Both caspofungin and micafungin have been shown to be not inferior to lipid formulations of amphotericin B for the treatment of invasive candidiasis. In addition, they have been found to have fewer adverse effects, especially those effects that would necessitate discontinuance of therapy (e.g., renal toxicity). One study has found anidulafungin to be superior to fluconazole.¹³⁹

Overall, the echinocandins are well tolerated. They are all associated with essentially the same set of adverse effects, which on the whole are not serious. Phlebitis, fever, nausea and vomiting are among the most commonly reported adverse effects. Thrombocytopenia, hypokalemia, and abnormal liver function test results are occasionally reported.

In susceptible patients, Candida species can enter the bloodstream either through translocation across the GI mucosa or via an intravascular catheter. This invasion occurs through three general pathways. The first such pathway is phagocytosis. Ingestion by phagocytes brings Candida into an intracellular position, and the leuckocytes then move across the endothelial cell lining and into the bloodstream. Candida species have indeed been observed within circulating leukocytes; however, the observation that fungal infections are common in neutropenic patients argues that a mechanism other than phagocytosis must also play a major role. The second pathway is passage of Candida pathogens between endothelial cells and thence into the bloodstream. The fenestrated endothelium in the kidney is one site at which this process might occur. The final pathway is endocytosis by the endothelial cells themselves. The process of endocytosis requires intact endothelial cell microfilaments and microtubules and is in part governed by tyrosine phosphorylation of endothelial cell proteins. Both killed and live Candida are ingested by this mechanism. In vitro studies have shpown that Candida hyphae induce their own endocytosis by expressing the invasinlike protein Als3, which binds to N-cadherin on the cell surface.¹⁴⁴ This binding then induces phosphorylation,¹⁴⁵ causing rearrangement of the microfilaments to produce pseudopods and initiate endocytosis.

It should be kept in mind that this is a highly simplified explanation and does not detail the many unknown specifics of the cellular processes involved. How Candida blastospores like those seen with C. glabrata gain access to the bloodstream is currently unknown.

1. Pfaller MA, Dickema DJ: Epidemiology of invasive candidiasis: a persistent public health problem. Clin Microbiology Rev 20:133, 2007

2. Pittet D, Monod M, Suter PM, et al: Candida colonization and subsequent infections in critically ill surgical patients. Ann Surg 220:751, 1994 [PMID 7986142]

3. Pelz R, Hendrix C, Swoboda S, et al: A double blind placebo controlled trail of prophylactic fluconazole to prevent Candida infections in critically ill surgical patients. Ann Surg 233:542, 2001 [PMID 11303137]

4. Voss A, Hollis RJ, Pfaller MA, et al: Investigation of the sequence of colonization and candidemia in nonneutropenic patients. J Clin Microbiol 32:975, 1994 [PMID 8027353]

5. Petri MG, Konig J, Moecke HP, et al: Epidemiology of invasive mycosis in ICU patients: a prospective multicenter study in 435 non-neutropenic patients. Paul-Ehrlich Society for Chemotherapy, Divisions of Mycology and Pneumonia Research. Intensive Care Med 23:317, 1997 [PMID 9083235]

6. Pelz R, Lipsett PA, Swoboda SM, et al: The diagnostic value of fungal surveillance cultures in critically ill patients. Surg Infections 1:273, 2000

7. Charles PE, Dalle F, Aube H, et al: Candida spp. colonization significance in critically ill medical patients: a prospective study. Intensive Care Med 31:393, 2005 [PMID 15711782]

8. Piarroux R, Grenouillet F, Balvay P, et al: Assessment of pre-emptive treatment to prevent severe candidiasis in critically ill surgical patients. Crit Care Med 32:2443, 2004 [PMID 15599149]

9. Pittet D, Monod M, Suter PM, et al: Candida colonization and subsequent infections in critically ill surgical patients. Ann Surg 220:751, 1994 [PMID 7986142]

10. Ascioglu S, Rex JH, DePauw B, et al: Defining opportunistic invasive fungal infections in immunocompromised patients with cancer in hematopoietic stem cell transplants: an international consensus. Clin Infect Dis 34:7, 2002 [PMID 11731939]

11. Vincent JL, Anaissie E, Bruining H, et al: Epidemiology, diagnosis and treatment of systemic Candida infection in surgical patients under intensive care. Intensive Care Med 24:206, 1998 [PMID 9565801]

12. Eggimann P, Garbin J, Pittet D: Epidemiology of Candida species infections in critically ill non-immunosuppressed patients. Lancet Infect Dis 3:685, 2003 [PMID 14592598]

13. Samonis G, Gikas A, Anaissie EJ, et al: Prospective evaluation of effects of broad spectrum antibiotics on gastrointestinal yeast colonization of humans. Antimicrob Agents Chemother 37:51, 1993 [PMID 8431017]

14. Ostrosky-Zeichner L, Sable C, Sobel J, et al: Multicenter retrospective development and validation rule for nosocomial invasive candidiasis in the intensive care setting. Eur J Clin Micribiol Infect Dis 26:271, 2007

15. Leon C, Ruiz-Santana S, Saavedra P, et al: A bedside scoring system ('Candida score') for early antifungal treatment in nonneutropenic critically ill patients with Candida colonization. Crit Care Med 34:730, 2006

16. Pappas PG, Rex JH, Sobel JD, et al: Guidelines for treatment of candidiasis. Clin Infect Dis 38:161, 2004 [PMID 14699449]

17. Rex JH, Sobel JD: Prophylactic antifungal therapy in the intensive care unit. Clin Infect Dis 32:1191, 2001 [PMID 11283809]

18. Lipsett PA: Clinical trials of antifungal prophylaxis among patients in surgical intensive care units: concepts and considerations. Clin Infect Dis 39:S193, 2004 [PMID 15546117]

19. Bernhardt HE, Orlando JC, Benfield JR, et al: Disseminated candidiasis in surgical patients. Surg Gynecol Obstet 134:819, 1972 [PMID 5031497]

20. Schwesinger G, Junghans D, Schroder G, et al: Candidosis and aspergillosis as autopsy findings from 1994 to 2003. Mycoses 48:176, 2005 [PMID 15842333]

21. Martin GS, Mannino DM, Eaton S, et al: The epidemiology of sepsis in the United States from 1979 through 2000. N Engl J Med 348:1546, 2003 [PMID 12700374]

22. Beck-Sague CM, Jarvis W, and the National Nosocomial Infections Surveillance System Secular trends in the epidemiology of nosocomial fungal infections in the United States, 1980–1990. J Infect Dis 167:1247, 1993

23. Centers for Disease Control National Nosocomial Infections Surveillance (NNIS) System Report, data summary from January 1992 through June 2003, issued August 2003. Am J Infect Control 31:481, 2003

24. Wisplinghoff H, Bischoff T, Tallent SM, et al: Nosocomial bloodstream infections in US hospitals: analysis of 24,179 cases from a prospective nationwide surveillance study. Clin Infect Dis 39:309, 2004 [PMID 15306996]

25. Kao AS, Brandt ME, Pruitt WR, et al: The epidemiology of candidemia in two United States cities: results of a population-based active surveillance. Clin Infect Dis 29:1164, 1999 [PMID 10524958]

26. Blumberg HM, Jarvis WR, Soucie JM, et al: Risk factors for candidal bloodstream infections in surgical intensive care unit patients: the NEMIS prospective multicenter study. The National Epidemiology of Mycosis Survey. Clin Infect Dis 33:177, 2001 [PMID 11418877]

27. McNeil MM, Nash SL, Hajjeh RA, et al: Trends in mortality due to invasive mycotic diseases in the United States, 1980–1997. Clin Infect Dis 33:641, 2001

28. Rentz AM, Halpern MT, et al: The impact of candidemia on length of hospital stay, outcome, and overall cost of illness. Clin Infect Dis 27:781, 1998 [PMID 9798034]

29. Pelz R, Lipsett PA, Swoboda SM, et al: Candida infections: outcome and attributable ICU costs in critically ill patients. J Intensive Care Med 15:255, 2000

30. Goff DA, Sierawsk SJ, Fass RJ: Cost analysis of Candida infection among surgical intensive care unit patients. Clin Drug Invest 12:176, 1996

31. Wey SB, Mori M, Pfaller MA, et al: Hospital-acquired candidemia: the attributable mortality and excess length of stay. Arch Intern Med 148:2642, 1988 [