Risk of embolic events before and after antibiotic treatment initiation among patients with left-side infective endocarditis

Purpose Embolic events (EEs) are a common complication of left-side infective endocarditis (IE). The aim of the present study was to identify risk factors for the occurrence of EEs before or after antibiotic treatment instauration among patients with definite or possible IE. Methods This retro-prospective study was conducted at the Lausanne University Hospital, Lausanne, Switzerland, from January 2014 to June 2022. EEs and IE were defined according to modified Duke criteria. Results A total of 441 left-side IE episodes were included (334: 76% were definite and 107; 24% possible IE). EE were diagnosed in 260 (59%) episodes; in 190 (43%) before antibiotic treatment initiation and 148 (34%) after. Central nervous system (184; 42%) was the most common site of EE. Multivariable analysis identified S. aureus (P 0.022), immunological phenomena (P < 0.001), sepsis (P 0.027), vegetation size ≥ 10 mm (P 0.003) and intracardiac abscess (P 0.022) as predictors of EEs before antibiotic treatment initiation. For EEs after antibiotic treatment initiation, multivariable analysis revealed vegetation size ≥ 10 mm (P < 0.001), intracardiac abscess (P 0.035) and prior EE (P 0.042), as independent predictors of EEs, while valve surgery (P < 0.001) was associated with lower risk for EEs. Conclusions We reported a high percentage of EEs among patients with left-side IE; vegetation size, intracardiac abscess, S. aureus and sepsis were independently associated with the occurrence of EEs. In addition to antibiotic treatment, early surgery led to further decrease in EEs incidence.


Introduction
Although infective endocarditis (IE) is a rare disease with 1.5-11.6cases per 100,000 people [1], it remains a life-threatening condition associated with an in-hospital mortality ranging from 15% to 30% and an 1-year mortality approaching 50% [2].Embolic events (EEs) are one of the most common complications (21-50%) associated with IE and have a great influence on patients' outcome [3][4][5][6][7].A decrease in the EE's incidence after antibiotic treatment initiation was previously described in several previous studies.However, most of these studies were performed more than two decades ago, when imaging for EEs detection was less frequently performed; therefore, asymptomatic EEs were less likely to be detected [3,4,[8][9][10].In addition, several studies also accepted diagnosis of EEs (such as stroke) only based on clinical symptoms/signs, without formal confirmation by imaging [4,8,10].
Many predictors of EEs have been previously identified, the most important among them being vegetation size > 10 mm [2,4,6,7,11,12], S. aureus IE [6][7][8]11], and prior EEs [4,13,14].Early recognition of such predictors might help clinicians to identify patients, who could benefit from early surgery in order to prevent further embolization.To date, only the size of vegetation with or without severe valvular dysfunction and prior EEs are recognized surgical indications to prevent further EEs according to from the European Society of Cardiology (ESC) guidelines, since previous studies have shown a beneficial effect of early surgery on occurrence of EEs [15,16].
The aim of the present study was to identify risk factors associated with the occurrence of EEs both before and after antibiotic treatment instauration.

Study design
This study was conducted at the Lausanne University Hospital, Lausanne, Switzerland, a 1100-bed primary and tertiary care hospital from January 2014 to June 2022 (2014-17: retrospective cohort; 2018 onwards: prospective cohort).

Patients
Inclusion criteria were adult patients (≥ 18 years old) and left-side IE according to modified Duke criteria.Additional inclusion criterion for the prospective cohort was the written consent and for the retrospective cohort the absence of refusal of the use of their data.A subsequent episode was excluded if it occurred within two months from the initial one.
Data regarding demographics (age, sex), comorbidities, cardiac predisposing factors [15], cardiac implantable electronic devices (CIEDs), microbiologic etiology, systemic symptoms, fever, acute heart failure, sepsis or septic shock, heart murmur, immunological phenomena [15], site of cardiac involvement and type of lesion (according to cardiac imaging modalities, macroscopic lesions on surgery or autopsy), cardiac surgery (timing), results of thoracoabdominal and cerebral imaging studies, embolic events (type, timing, symptoms) were retrieved from patients' electronic health records.Study data were collected and managed using REDCap by an infectious diseases specialist.REDCap electronic data capture tools is hosted at Lausanne University Hospital.REDCap (Research Electronic Data Capture) is a secure, web-based software platform designed to support data capture for research studies [17,18].

Management of IE
According to internal guidelines, an infectious diseases consultation with a thorough physical examination was performed on a mandatory basis for all patients with IE suspicion.Thoraco-abdominal (computed tomography, abdominal magnetic resonance imaging or 18 F-fluorodeoxyglucose positron emission positron emission tomography-computed tomography) and cerebral imaging (computed tomography or magnetic resonance imaging) were performed in all patients with clinical suspicion of EE based on local symptoms.Their realization in asymptomatic patients was left at the discretion of the treating physician and infectious diseases consultant.An endocarditis-team was established on January 2018, comprising of infectious diseases specialists, cardiologists, cardiac surgeons, which reviewed all patients with IE suspicion during weekly meetings.

Definitions
The definition of EEs included major peripheral artery embolism, septic lung emboli, hepatic, renal or splenic emboli, mycotic aneurysm, ischemic or hemorrhagic stroke, cerebral abscess, conjunctival bleeding, retinal emboli, chorioretinitis, Janeway lesions or nail bed bleeding.EEs were divided according to their timing of occurrence into EE at those presented before versus after administration of antibiotic therapy.EEs were considered symptomatic if the patient presented local symptoms, such as confusion, headache, seizures, neurologic deficit for central nervous system EE, dyspnea, thoracic pain, cough for intrathoracic EE or abdominal pain, back pain for intraabdominal EE.Cutaneous EE (Janeway lesions or nail bed hemorrhages) were considered asymptomatic.For symptomatic EE, the date of EE was defined as the date of symptoms' onset attributed to EE as reported by the patient.
IE was defined according to the ESC modified Duke criteria [15].IE was characterized as community, healthcare or nosocomial according to Friedman et al. [19] Infection was categorized as sepsis or septic shock according to definition proposed by the Sepsis-3 International Consensus [20].
Valvular surgery within after antibiotic treatment initiation was included.A subset of patients that benefited from surgery but only if it was performed before the occurrence of EE after antibiotic treatment initiation was also included.

Endpoint
The primary endpoint was incidence of EEs occurring within two months after the initiation of antibiotic therapy.Patients were followed until two months after antibiotic initiation (medical records review or telephone call) or death.

Analysis
SPSS version 26.0 (SPSS, Chicago, IL, USA) software was used for data analysis.Categorical variables were analyzed using the chi-square or Fisher exact test and continuous variables with Mann-Whitney U test.Variables in bivariate analyses with P < 0.1 that did not contribute to multicollinearity were entered into the multivariable analyses.After checking Cox assumptions, two multivariable Cox proportional hazards (PH) regression models were performed with dependent variables being overall EEs and EEs occurring after antibiotic administration; for both models, valve surgery was treated as a time dependent covariable.Bivariate multivariable logistic regression analysis was performed with dependent variable being EEs occurring before antibiotic administration.Adjusted odds ratios (aORs) and 95% confidence intervals (CIs) were calculated to evaluate the strength of any association.All statistic tests were twotailed and P < 0.05 was considered statistically significant.Kaplan-Meier curve of the embolic event probability after 4 days on antibiotic treatment of patients with left-side IE was performed for patients in order to assess the role of valvular surgery within 4 days after antibiotic treatment instauration.

Site of embolic events
Central nervous system (184; 42%) was the most common site of EE, followed by the spleen (72; 16%).Table 1 shows the site of EEs before and after antibiotic treatment initiation.A thoracoabdominal imaging study was performed in 353 (80%) episodes (in the absence of symptoms in 196; 56%) and a cerebral imaging in 299 (68%) episodes (in the absence of symptoms in 138; 46%) (Table 2).EEs discovery by thoracoabdominal imaging studies did not differ among patients with and without symptoms (47% versus 45%; P 0.731).Cerebral imaging studies were more prone to detect EEs in symptomatic patients than in asymptomatic ones (68% versus 31%; P < 0.001).
Table 4 summarizes the factors associated with new EE identified after antibiotic treatment initiation.Cox PH regression model (Table 4) revealed EE before antibiotic treatment initiation (P 0.042; OR 1.43, CI 1.13-2.02),sepsis (P 0.009; OR 1.56, CI 1.12-2.16),vegetation size ≥ 10 mm (P 0.005; OR 1.67, CI 1.17-2.39)and intracardiac abscess   Figure 2 shows a Kaplan-Meier curve of the embolic event probability after 4 days on antibiotic treatment of patients with left-side IE according to valvular surgery performed within 4 days after antibiotic treatment initiation; valvular surgery was associated with lower risk of embolic events (P 0.005).
As previously reported, EE involved more commonly the central nervous system [3, 5-7, 22, 23].The incidence of symptomatic central nervous system EEs (24%), were higher than previously reported (10-15%) [3-6, 8, 22].A possible explanation could be that in our center a cerebral imaging study was systematically performed for any central nervous system symptom such as neurologic deficit, seizures, confusion or headache, while in previous studies,  imaging studies were performed only in patients with a neurologic deficit [8].After the antibiotic treatment, new symptomatic central nervous system EE occurred in 16%, which was also higher than previously reported (4-6%) [7,8].
Previous studies showed an association between S. aureus, the most common cause of IE, and the risk of EEs [6-8, 11, 24].In our study, S. aureus was associated with EE before antibiotic treatment initiation, but it did not influence the EEs' risk after antibiotic initiation; this discordance was previously found in other studies [7].
Likewise, many studies have shown that vegetation's size was a major determinant of EE risk [2,4,6,7,11,12].In the present study, vegetation size ≥ 10 mm was associated with the occurrence of EE both before and after antibiotic treatment initiation.We also found that the presence of intracardiac abscess was also associated with EE occurrence before or after antibiotic treatment initiation, a finding which was not observed in the meta-analysis, where intracardiac lesions other than vegetation had no significant influence on EE risk [11].Intracardiac abscess was found to be associated with central nervous system EE after the initiation of antibiotic treatment in a study from the International Collaboration on Endocarditis Prospective Cohort Study [8].Finally, even though mitral valve endocarditis was previously shown to be associated with high risk for EE [8,11], this variable failed to achieve statistical significance in our multivariable model.
EEs at presentation were found to predispose to EEs after the initiation of antibiotic treatment, as previously reported [4,13,14].Their importance was underlined by the fact that EEs at initial presentation are part of the risk prediction score for EEs after antibiotic treatment initiation [12].Prior EEs in association with vegetation size > 10 mm is a recognized indication for valve surgery according to the ESC guidelines for the prevention of further embolism [15].In the present study, a reduction in EEs was achieved with early surgery.While the performance of early surgery in patients with an operative indication was found in a meta-analysis to be associated with lower in-hospital and 1-year mortality as compared to patients treated with antimicrobial treatment only or antimicrobial treatment and late operation, the role of early surgery for the reduction of EE risk remains unclear [25].The present study reinforces the role of early surgical management in patients with operative indication, in order to reduce the risk of further EEs, since 42% of patients with a vegetation > 10 mm developed an embolic event after antibiotic treatment initiation [16,26,27].In Data are depicted as number/percentage or median/Q1-Q3 CIED cardiac implantable electronic devices; CRP C-reactive protein; HACEK Haemophilus spp., Aggregatibacter spp., Cardiobacterium hominis, Eikenella corrodens, Kingella kingae; IE infective endocarditis   In the present study, sepsis was associated with EEs before antibiotic treatment initiation.To the best of our knowledge, this is the first study to show such an association.A possible explanation could be that EE and sepsis are part of more severe IE, caused by S. aureus.Indeed, in the present study sepsis was more common in patients with S. aureus IE, also being a predictor of EE.It was previously shown that patients with sepsis (qSOFA score ≥ 2) were at higher risk for adverse events, including EE [28].
The study has several limitations.First, it was monocentric, with almost one third of patients being retrospectively collected.Another limitation was that vegetation motility, that was previously found to be associated with EEs, was not evaluated in the present study [11].Moreover, 19% of patients did not have a TOE, thus the calculation of the vegetation size was only based only on the TTE in those patients.A referral bias applied to the present study, since our center was the referral center for cardiac surgery.In addition, not all patients benefited from cerebral or cerebral imaging studies, thus the incidence of asymptomatic EE may have been underestimated [23].Furthermore, we cannot exclude that the etiology of some EEs was other than IE (p.ex.atrial fibrillation).
In conclusion, we reported a high percentage of EEs (59%) among patients with left-side IE.Vegetation size, intracardiac abscess, IE due to S. aureus and sepsis were associated with occurrence of EEs.Even though EEs risk declined steadily during treatment, EEs remained a frequent occurrence, especially in patients with prior EEs.In addition to antibiotic treatment, early surgery, led to further decrease in EEs incidence.

Fig. 1
Fig. 1 Timing of embolic events in relation to antibiotic treatment initiation

Fig. 2
Fig. 2 Kaplan-Meier curve of the embolic event probability after 4 days on antibiotic treatment of patients with left-side IE according to early (within 4 days after antibiotic treatment initiation) valve surgery

Table 1
Site of embolic events before and after antibiotic treatment initiation in 441 patients with left-side infective endocarditis

Table 2
Predictors of all embolic events (before and after antibiotic treatment initiation) and results of the Cox PH multivariable regression

Table 3
Predictors of embolic events before antibiotic treatment initiation and results of the multivariable logistic regression

Table 4
Predictors of embolic events after antibiotic treatment initiation and results of the Cox PH multivariable regression independently of the timing of the embolic event.Since the risk of EE during antibiotic treatment is higher in the two first weeks, surgery in patients with a surgical indication for embolism prevention, could be more beneficial if Data are depicted as number/percentage or median/Q1-Q3 CIED cardiac implantable electronic devices; CRP C-reactive protein; HACEK Haemophilus spp., Aggregatibacter spp., Cardiobacterium hominis, Eikenella corrodens, Kingella kingae; IE infective endocarditis; PH proportional hazard a Treated time dependent covariable