Introduction

Candida infective endocarditis (IE) is a rare and poorly understood complication of fungemia. Although Candida IE has been regarded traditionally as an uncommon infection, the rates of fungemia have increased by as much as 128% in recent years, leaving a growing number of patients at risk for this complication [1]. Despite aggressive antifungal and surgical therapy, mortality approaches 80% in some series and a better understanding of this infection is needed [2–4].

Because of the rarity of candidal IE at any single institution, the epidemiology, prognosis, and optimal therapy of Candida IE are poorly defined, and treatment guidelines are derived mostly from single-site case series and case reports [3–6]. The recommended treatment of Candida IE is an amphotericin B-based regimen plus surgical intervention, often followed by long-term fluconazole for suppression [5]. However, because of the availability of safe, effective drugs for invasive candidiasis, emerging azole resistance, and high mortality, alternative drugs are now being increasingly used for Candida IE [7–14].

In the current investigation, we used a contemporary, prospective, international, multi-center cohort of patients with definite endocarditis to better evaluate the clinical characteristics, current antifungal treatment practices, and outcome of patients with Candida IE. Moreover, we compare and contrast Candida IE cases with non-fungal cases in order to highlight differences in epidemiology and outcomes.

Materials and methods

Study population

The patient data are derived from the International Collaboration of Endocarditis—Prospective Cohort Study (ICE-PCS), a multi-national database of prospective cases of endocarditis. Details of the ICE-PCS have been described previously [15–17]. From June 2000 to August 2005, there were 2,760 cases of definite IE contributed by 61 centers in 28 countries. Of the 2,760 cases of definite IE, there were 33 cases due to Candida spp. All cases were classified as definite IE based on revised Duke criteria [18] and all cases were verified by the coordinating center (Duke University Medical Center). Fungal IE cases caused by organisms other than Candida (11 cases) were excluded from analysis. From each enrolled patient, data were collected from the index hospitalization and entered using an Internet-based system. The data collected included demographics, symptoms associated with IE, underlying medical conditions, predisposing factors, clinical signs and symptoms, antifungal therapy, echocardiographic findings, associated complications, and outcomes (stroke, embolic events, heart failure, intracardiac abscesses, persistently positive blood cultures, and death). Healthcare-associated IE was defined as either nosocomial infection or non-nosocomial healthcare-related infection. Nosocomial infection was defined as IE developing in a patient hospitalized for more than 48 h before the onset of signs/symptoms consistent with IE. Death was determined at the time of hospital discharge. Data on longer-term mortality was not collected.

Statistical methods

Categorical variables were represented as frequencies and percentages of the specified group. The associations between clinical characteristics and Candida IE were measured using the Wilcoxon rank sum test for continuous variables and Chi-square or Fisher’s exact methods for categorical variables. For all tests, statistical significance was determined at the 0.05 level. All statistical analyses were performed using SAS software (version 9.1, SAS Institute, Cary, NC).

Results

Patient characteristics

Of the 2,749 patients with definite IE, 33 (1.2%) were Candida IE cases. The mean age of patients with Candida IE was 54.9 years. Patient characteristics including diabetes, renal disease, malignancy, intravenous drug use, and congenital heart disease were similar between the two groups (Table 1). Patients with Candida IE were less likely to be male (51.5% vs. 67.9%, p = 0.04), more frequently had previous endocarditis (21.2% vs. 7.8%, p = 0.005), and were more likely to have short-term indwelling catheters (21.2% vs. 4.4%, p < 0.0001). Among patients who had an invasive procedure within 60 days prior to the onset of symptoms, coronary artery bypass grafting (CABG) was more common among Candida IE patients (22.2% vs. 3.7%, p = 0.007). Prosthetic valve IE was more common in Candida patients (48.8% vs. 19.6%, p = 0.005), and Candida IE patients were more likely to have the infection classified as being healthcare-related (51.5% vs. 25.8%, p = 0.0009).

Table 1 Characteristics of patients with Candida and non-fungal endocarditis from the International Collaboration of Endocarditis (ICE) database (n = 2,749)
Full size table

Clinical findings

Of patients with any IE etiology, most (75%; 2,068/2,749) experienced the first clinical manifestation less than one month before presentation, and the timing of IE manifestations was similar between the two groups. The most common clinical manifestations among all of the patients were fever (79.5%; 2,170/2,728), new murmur (47.9%; 1,053/2,198), hematuria (22.1%; 607/2,737), pulmonary edema (22.3%; 556/2,491), and evidence of a vascular embolic event (15.9%; 435/2,728). Overall, there was little difference in symptoms and signs at presentation between the Candida and non-fungal IE groups (Table 2). A total of 1,316 (47.9%) of 2,749 patients had surgery for endocarditis, and this was not different for the two groups. Candida IE patients were more likely to have surgery indicated because of embolization (40% vs. 19.8%, p = 0.054), persistent fungemia (33% vs. 9.9%, p = 0.003), and myocardial abscess (46.7% vs. 22.2% p = 0.026). By contrast, surgery for the indications of congestive heart failure (42.6% vs. 13.3%, p = 0.02) and valvular regurgitation (68% vs. 40%, p = 0.018) were more common in patients with non-fungal IE.

Table 2 Clinical findings of patients with Candida and non-fungal endocarditis
Full size table

Complications

Congestive heart failure, systemic embolization after presentation, and stroke were common but had similar in occurrence in the two groups. Candida IE was associated with persistently positive blood cultures (39.4% vs. 8.8%, p < 0.001) (Table 3). Mortality at the time of discharge was higher among Candida IE patients than non-fungal IE patients (30.3% vs. 17%, p = 0.046). This mortality difference was more pronounced among those patients who had surgery for this episode of IE (33.3% vs. 13.8%, p = 0.03). Among 15 Candida IE patients who underwent surgical intervention for this episode of endocarditis, mortality at discharge was similar to Candida IE patients who did not have surgery (33.3% vs. 27.8%, p = 0.26). Those patients who underwent surgical intervention were more likely to have previous IE (40% vs. 5.7%, p = 0.016), previous surgery for IE (33.3% vs. 5.6%, p = 0.009), paravalvular complications on ECHO (46.7% vs. 11.1%, p = 0.015), and systemic embolization (46.7% vs. 16.7%, p = 0.04) when compared with patients with Candida IE who were not treated with surgical intervention.

Table 3 Complications and outcomes of patients with Candida and non-fungal endocarditis
Full size table

Organisms and antifungal treatment

Among the 33 patients with Candida IE, 16 (48%) were caused by C. albicans, 7 (21%) C. parapsilosis, 5 (15%) C. glabrata, and 3 (9%) C. tropicalis. Two (6%) isolates were not fully speciated. Treatment data were available for 27 (82%) of 33 patients (Table 4). The most common antifungal agent used was amphotericin B (AmB), either conventional AmB (13/27; 48.1%) or a lipid formulation (3/27; 11.1%). Fluconazole was used in 12 (44.4%) of 27 patients. Primary therapy with fluconazole was used in 6 (54.5%) of 11 patients with complete fluconazole treatment data available. Ten patients (37%) received treatment with the newer antifungal agents caspofungin or voriconazole. Among the patients who received single-drug therapy, death occurred in 6 (40%) of 15 patients; death occurred in 2 (25%) of 8 who received sequential therapy. In only two cases, combination therapy was used and both patients were alive at discharge. Two (20%) of 10 people who received newer therapies (caspofungin or voriconazole) died.

Table 4 Treatment for 27 patients with Candida infective endocarditis (IE)
Full size table

Discussion

Candida IE is an uncommon but frequently fatal infection [3, 4, 6]. A better understanding of the epidemiology, associated risk factors, and treatment methods is needed, but it is difficult to obtain because of the rarity of cases and the lack of large prospective cohorts. We compared contemporary clinically well-characterized cases of candidal IE to non-fungal IE cases registered as part of a large, multi-center, prospective dataset to better understand Candida IE. This analysis revealed several important observations regarding predisposing conditions, clinical findings, and treatment modalities.

Important risk factors or predisposing conditions for fungal endocarditis have been reported in recent, extensive reviews, and the most frequently reported are previous surgery, vascular lines, antibiotic use, underlying heart disease, prosthetic valves, and immunocompromising conditions [2–4, 6] . We found similar predisposing conditions and noted several distinct differences among Candida and non-fungal IE cases. First, CABG and prosthetic valve IE were significantly more common in Candida patients. An increase in previous CABG among Candida IE patients could be explained by CABG being performed in association with prosthetic valve surgery. Second, healthcare-associated IE was more common among patients with Candida IE. The increase in hospital-acquired Candida IE, in general, is consistent with recent data describing Candida as an emerging nosocomial bloodstream pathogen over the past decade [19].

The clinical findings and presentation of patients with Candida and non-fungal IE are very similar, as has been previously described [6]. The most important exceptions discovered in our review are related to indications for cardiac surgery. Of patients who had surgery during this episode of IE, those with Candida IE were more likely to have surgery based on the finding of myocardial abscess or persistently positive blood cultures. Non-fungal cases more commonly had heart failure or valvular insufficiency as a reason for surgery.

There were few differences in complications and outcomes in the two groups except mortality. Candida IE mortality has been reported to be up to 80% in previous reviews [2–4, 6], but variability in the data collection and description of individual cases makes it difficult to determine an appropriate risk of death. Ellis et al. [3], in a recent review, demonstrated that the crude survival of patients with fungal endocarditis had increased over the past 20 years, from 14% before 1970 to 41% in the period 1991–1995. Possible reasons for this improved survival were attributed to better echocardiographic techniques, earlier diagnosis of endocarditis, or better supportive care of ill patients [3]. Nearly one-third of patients in our series died during hospitalization, with mortality significantly greater than non-fungal cases. The mortality among patients with Candida IE in our series is surprisingly less than that reported in previous reviews, but may be due to a multitude of factors. Diagnostic and treatment modalities have improved in the past decade, but, likely, cannot account for such a difference in survival. The inclusion of Candida cases only, which often have better survival compared to other fungal causes [3, 4], and the survival end-point defined at hospital discharge (compared to literature reviews, where follow-up data were available for up to several years) may reflect the lower mortality in this series [3]. Finally, the use of newer antifungal therapies, such as the echinocandins and lipid preparations of amphotericin B, not included in previous reviews because of the lack of availability, may have an impact on outcomes and warrant further evaluation.

The traditional antifungal treatment of Candida IE is amphotericin B (6–8 weeks), often followed by fluconazole as suppression because of frequent relapse [5, 6]. In addition, surgical intervention with valve replacement is generally recommended in most cases. The combination of antifungal and surgical therapy is purported to be more beneficial than antifungal therapy alone, although controlled studies have not been performed for confirmation [3, 4, 20]. In this cohort, surgical therapy was not associated with increased survival compared to antifungal therapy alone. It is encouraging that patients who did not receive surgical therapy fared relatively well; however, we speculate that the lack of a significant difference between the groups may reflect a combination of factors, including increased morbidity and complications at presentation among patients who underwent surgery. Patients who underwent surgical intervention were more likely to have previous IE, previous surgery for IE, paravalvular complications on ECHO, and systemic embolization. Although these may be important differences that influenced the risk of death, with the limited number of patients evaluated, it is difficult to draw conclusions with respect to the appropriate management.

In this cohort, an amphotericin B preparation was the most frequent drug used. Fluconazole was the second most common, and was used either for primary or sequential therapy. Sequential therapy was frequently employed, and mortality in this group was lower than in patients who received a single agent. This probably results from selecting a subset of patients that lived long enough to “step down” to azole therapy. The lengths of therapy and dosages were not captured, so appropriate comparisons cannot be made. An important obstacle in the successful antifungal therapy of Candida IE has been adverse events associated with prolonged amphotericin B administration. With the approval of new antifungal agents in the past several years, specifically echinocandins and newer azoles, questions have arisen about the role of these agents for the treatment of Candida IE. The echinocandins and voriconazole have shown efficacy and safety for the treatment of invasive candidiasis and candidemia [21, 22]; however, data on usage in endocarditis is limited to case reports [7–14]. Although some clinical success has been documented, selection bias may be present, and determinations of efficacy cannot be made. Our series reflects a shift in the treatment of Candida IE. More than one-third of patients received newer antifungal agents, particularly the echinocandin, caspofungin, and mortality among these patients (20%) was similar to the other groups. Adverse events from drug use and isolate susceptibilities were not captured in the database, so the reasons for the use of these drugs are unclear.

Although an important aspect of this dataset is its overall size, and this represents the largest reported number of definite Candida IE cases compared to non-fungal cases, there are important limitations. The data were collected prospectively, but analysis was conducted retrospectively. The number of Candida cases is not large enough to draw conclusions regarding treatment, and long-term mortality data were not collected.

These data represent a multi-center collaborative effort describing a large cohort of definite endocarditis cases. There appear to be distinct epidemiologic features of Candida IE when compared to non-fungal cases. Indications for surgical intervention are different, mortality is increased, and alternative antifungal treatment options are increasingly used for this devastating disease. Large datasets or series, despite their limitations, are needed to help better define Candida IE.