Introduction

Current guidelines recommend systematic screening with echocardiography in all patients with clinically suspected infective endocarditis (IE), but also in patients with bacterial bloodstream infections (BSI) caused by Gram-positive bacteria, particularly those caused by Staphylococcus aureus, irrespective of a clinical suspicion of endocarditis [1, 2]. Echocardiography in that population frequently demonstrates the presence of clinically unsuspected IE.

The recommendation to apply echocardiography systematically in other BSI is far from clear. The incidence of Candida infective endocarditis (CIE) in candidemic patients has been evaluated only in a retrospective study by Nasser et al. [3], who analyzed patients with prosthetic heart valves. We were unable to find studies investigating the yield of a systematic use of echocardiography to rule CIE in patients with candidemia [4]. The incidence of clinically unsuspected CIE is unknown.

We performed a prospective study to assess the yield of routine echocardiography in an unselected cohort of adult patients with candidemia.

Patients and methods

Our institution is a 1,550-bed tertiary teaching hospital. During the study period, its catchment population was 650,000–750,000 inhabitants. The hospital is a referral center with an active major heart surgery unit and several transplantation programs. Our institution also has a cooperative multidisciplinary group for the prospective study of IE (GAME, Grupo de Apoyo al Manejo de la Endocarditis).

This study was reviewed and approved by the Ethics Committee of the Hospital Universitario Gregorio Marañón.

Study period and patient selection

The GAME endocarditis study group has systematically followed all patients with candidemia at our institution since 2007. A microbiological endocarditis alert is activated when the microbiology department detects a patient with candidemia, and a specialized nurse and an infectious diseases specialist evaluate the patient’s clinical condition and routinely recommend transesophageal echocardiography (TEE) to the attending physician and the patient. The recommendation is made independently of the clinical suspicion of endocarditis, which is based mainly on the presence of valve disease, prosthetic endovascular material, or persistent candidemia. Patients are followed up until discharge.

Data for all patients with candidemia between 2007 and October 2012 were registered in a BSI database and an IE database.

Echocardiography

The type of echocardiography was chosen by the cardiologist. Usually, a transthoracic echocardiography (TTE) was performed first, and then a TEE was considered. But when the suspicion of IE was high, some cardiologists proceeded directly to the TEE, due to its higher sensitivity. The routine recommendation to exclude endocarditis was TEE. Depending on whether consent was given or not for TEE, the patient’s clinical status, the presence of prosthetic material, and whether a good window was obtained with TTE, a TTE, a TEE, or both were done. Patients gave their informed consent for TEE. As TTE was considered a routine test, no formal signed consent was required.

Processing in the microbiology laboratory

Blood cultures were obtained using standard procedures and processed using the BACTEC 9240 blood culture system until 2009 (Becton Dickinson, Sparks, MD, USA) and the BD Bactec FX system from 2010 onwards (Becton Dickinson, Sparks, MD, USA). All systems were used according to the manufacturer’s instructions. Candida species were identified by ID32 (bioMérieux, Marcy L’Étoile, France).

Clinical criteria and definitions

We defined an episode of candidemia as the isolation from ≥1 blood culture of a microorganism belonging to the genus Candida. Candidemia was considered persistent when ≥1 blood culture obtained ≥3 days after the first one yielded the same Candida species.

We recorded the following clinical data: demographic characteristics, underlying diseases (in particular, valve disease), valvular prosthesis, persistent candidemia, and Candida species. In cases of confirmed CIE, we recorded the type of echocardiography performed, treatment, and outcome.

The diagnosis of CIE was made according to the modified Duke criteria [5]. Regarding the echocardiographical criteria, all patients had mobile masses compatible with endocardial vegetations. If the vegetations were not located on valves, but were mural vegetations, they should not be implanted in central catheters or should persist after withdrawal of the catheter in order to be considered diagnostic of endocarditis.

All cases of endocarditis were discussed at the GAME study group in order to evaluate if they fulfilled endocarditis criteria and, if so, they were included in the endocarditis database.

Statistical analysis

We analyzed the yield of echocardiography for the diagnosis of CIE, in particular that of TEE.

Associations between variables were evaluated using the χ2 test for categorical variables, the t-test for normally distributed continuous variables, and the Mann–Whitney test for non-parametric comparisons. A p-value < 0.05 was considered significant. Multivariate analysis was performed using logistic regression. Variables with a p-value ≤ 0.1 in the univariate analysis were included in the multivariate model.

The statistical analysis was performed with SPSS 16.0 (SPSS, Chicago, IL, USA).

Results

From January 2007 to October 2012, we recorded 263 episodes of candidemia in adult patients in our hospital, which represented 7.5 % of all BSI. Echocardiography was not performed in 76 cases and our results are based on data from the remaining 187 patients. Reasons for not performing echocardiography were as follows: patients had been discharged or had died when the blood culture became positive (17), patients were critically or terminally ill (38), and the patient or the physician refused the procedure (21). The only independent significant difference between patients that underwent an echocardiogram and those that did not was a higher prevalence of persistent candidemia among the former [33.7 % vs. 10.2 %, p < 0.01, relative risk (RR) 4.51 (1.83–11.11)]. None of the eligible patients who did not undergo echocardiography presented with IE or a recurrent episode of candidemia during the follow-up period.

Characteristics of the population

Table 1 shows the general characteristics of the population. Twelve patients (6.4 %) had a valvular prosthesis, 2 (1.1 %) had pacemakers, 26 (13.9 %) had non-prosthetic valve disease, 63 (33.7 %) had persistent candidemia, and 98 (52.4 %) had none of these factors. Fourteen patients presented >1 risk factor.

Table 1 General characteristics of the study population
Full size table

Type of echocardiogram

One or more echocardiograms were performed in 187 patients. Of these, 100 patients (53.5 %) underwent TTE only, 15 (8 %) underwent TEE only, and 72 patients (38.5 %) underwent both. The median time from blood culture positivity to echocardiography was 5 days [interquartile range (IQR), 3–7.5 days].

When comparing patients that underwent a TEE with those who did not, the only independent significant difference was the presence of valvular prosthesis, which was more prevalent among the former [12.6 % vs. 1 %, relative risk (RR) 13.6 (confidence interval [CI] 1.7–109.3)].

Yield of echocardiography

Of the 187 patients who underwent echocardiography, 11 (5.9 %) had findings that indicated IE. The diagnostic yield of IE in patients with candidemia was 2.9 % (5/172) for TTE and 11.5 % (10/87) for TEE. Left valves were the most commonly affected (five patients), followed by four cases of atrial endocarditis. C. albicans and C. parapsilosis were responsible of the majority of the endocarditis cases (8/11, 72.7 %) (Table 2).

Table 2 Characteristics of patients with Candida endocarditis
Full size table

Among the 11 patients with CIE, one had only a TTE performed, four had only a TEE performed, and six had both TTE and TEE. All patients presented mobile masses that fulfilled Duke’s criteria. CIE was diagnosed using TTE in 5 out of 7 cases of endocarditis in which it was performed (71.4 %); TEE was positive in all ten cases of CIE in which it was performed (100 %). Of the six patients with endocarditis who underwent both TTE and TEE, two cases of CIE (33.3 %) were diagnosed only using TEE.

The prevalence of CIE among patients with a valvular prosthesis was 33 % (4/12). In patients with at least one risk factor (valvular prosthesis, persistent candidemia, or previous valve disease), the prevalence was 9 %; in those without risk factors, it was 3.1 % (p = 0.12). Nevertheless, thanks to the use of routine echocardiography in patients with candidemia, CIE was diagnosed in three cases with neither endovascular predisposing conditions nor persistent candidemia (Table 2). Two of these patients presented mural vegetations that were not related with catheters or persisted after withdrawal, and, so, fulfilled Duke’s criteria.

Comparison between patients with and without endocarditis

In an attempt to determine whether echocardiography could be avoided in selected patients with a low risk of endocarditis, we compared patients with and without endocarditis who had undergone echocardiography. No statistically significant differences were observed between patients with or without CIE according to clinical or microbiological predisposing factors (Table 3), except for the presence of a valvular prosthesis, which was significantly more frequent among patients with IE (36.4 % vs. 4.5 %, p < 0.01). When the subset of patients with valvular prosthesis was excluded, it was not possible to predict which patients would have CIE.

Table 3 Comparison of patients with or without Candida infective endocarditis (CIE)
Full size table

Discussion

Routine echocardiography in patients with candidemia reveals a higher incidence of CIE than expected, particularly in patients examined using TEE (11.5 % of cases of CIE). As many as 30 % of cases of CIE were found in patients with candidemia who had neither persistent candidemia nor previous heart valve disease and were clinically unsuspected.

Several authors have underlined the role of clinical criteria and pretest probability in ruling out IE using echocardiography [6]. Unexplained bacteremia is considered to have a 5–40 % probability of being IE. The yield of routine echocardiogram in cases of BSI caused by S. aureus has been thoroughly studied [7–11], and current guidelines recommend that most patients with S. aureus bacteremia should be evaluated by echocardiography [2], even if they are classified as low risk [12], because the risk of endocarditis is as high as 13–22 %[11, 13].

CIE is an uncommon entity [14]. In their retrospective study, Nasser et al. suggested that patients with prosthetic heart valves who develop nosocomial candidemia are at notable risk of having or developing Candida prosthetic valve endocarditis months or years later [3]. To our knowledge, no previous studies have investigated prospectively the yield of echocardiography in candidemia.

Given the severity of and mortality associated with CIE, and the frequent delay of its diagnosis until significant vegetations or embolic complications are present, we consider that the prevalence of 11.5 % of cases of CIE among patients with candidemia studied with TEE is high enough to consider a recommendation for systematic echocardiography in this population. Considering the frequency of candidemia in our institution, this would only increase the workload of the echocardiography laboratory by less than four TEEs per month. Timely recognition enables better management and improved outcome in endocarditis [1, 15].

The only significant difference between patients with and without CIE was the presence of valvular prosthesis, which underlines the need to rule out CIE by TEE, especially in patients with a valvular prosthesis and candidemia. Except for this factor, we were not able to identify patients with a higher risk of IE.

In any case, the routine use of echocardiography confirmed three cases of CIE that would have gone undetected otherwise, because of the absence of valve disease, valvular prosthesis, or persistent candidemia. Atrial endocarditis, although uncommon, is well known [16–20], and in particular with Candida [21, 22]. As there were no clinical markers of CIE in this group, we recommend considering routine echocardiography for all patients with candidemia, including TEE when the TTE is not diagnostic of endocarditis, whenever the general situation of the patient allows it.

The limitations of our study are that, despite the intervention of the infectious diseases department, the number of patients with candidemia who did not undergo TEE was high, thus limiting the ability of the study to estimate the real prevalence of IE and introducing a possible selection bias. Although the recommendation was routine echocardiography, we found a significant difference between patients that underwent a TEE or not. TEE is not without risks, although serious complications are extremely rare, having been estimated at less than 1 in 5,000 [23]. Patients with candidemia are often too ill to undergo TEE, even when the approach is systematically recommended. Because of that, the actual incidence of CIE could be overestimated if there was a bias in performing TEE in some patients, which we cannot exclude. On the other hand, we cannot rule out the possibility that a diagnosis of CIE was missed because TEE was not performed systematically; therefore, our data could underestimate the real incidence of CIE among patients with candidemia. Ours is a single-institution study and, accordingly, interpretation of the results requires caution.

Nevertheless, systematic TTE or TEE in patients with candidemia reveals a significant proportion of episodes of unsuspected endocarditis and should be offered to all patients with candidemia.