Background and objectives
Infective endocarditis (IE) is a life-threatening diseases associated with high mortality and morbidity.
In case of suspected IE, the diagnosis should be made rapidly in order to offer an early antibiotic and
surgical treatment to the patient and improve prognosis. As cardiac imaging plays a critical role in the
diagnostic role of IE, this study aimed to assess the diagnostic accuracy of echocardiography in
suspected IE and to describe the typical echocardiographic lesions associated with this diagnosis. As a
secondary objective, the study investigated the clinical and imaging characteristics independently
associated with a final diagnosis of IE, or with the presence of a positive major echocardiographic
criterion according to the Duke Criteria for IE.
Methods
This retrospective single-center study included 300 consecutive adult patients referred to the
echocardiography laboratory of the CHUV for suspected IE between October 2016 and December
2017. Positive imaging was defined according to the 2015 guidelines of the European Society of
Cardiology. Endocarditis was considered present when the patient was treated for endocarditis as
mentioned in the discharge letter.
Results
Mean age was 64 years old and the majority were men (68%). Surgery was performed in 39% and inhospital
mortality was 13%. In IE+ patients (n=72, (24%)), the most frequent exam performed was
TTE (90%), followed by TEE (64%) and a small number of PET-CT (15%). TEE was less frequently
performed in prosthetic valve endocarditis (PVE) compared to native valve endocarditis (NVE) (55%
versus 67%), while PET-CT was more frequently performed in PVE compared to NVE.
TTE had an overall sensitivity and specificity of 60% and 95% respectively. After stratification
according to the type of heart valve, TTE has a lower diagnostic accuracy and especially a lower
sensitivity for PVE (Sensitivity 47%, Specificity 90%) compared to NVE (Sensitivity 65%, Specificity
95%). The overall sensitivity and specificity of TEE in this study were 80% and 95%, but it was as high
as 91% and 98% in the in the absence of valve prosthesis. In a small subgroup of patients with a
suspected PVE who underwent 18F-FDG PET-CT (n=9), the diagnostic accuracy of both TTE and TEE
was particularly low (Se 33% for both of them isolation, increasing to 67% in combination), but
increased significantly after adding 18F-FDG PET-CT (combined Se of 100%). Forty-six vegetations
were visualized in 44 IE patients (61%), with a median size of 14.5 [8.3-20.8] mm; 37 (71%) were
associated with NVE and 9 (45%) with PVE (p=0.04). Nine abscesses (13%) were visualized, 7 (13%)
on NVE and 2 (10%) on PVE (p=0.69). Ten perforations were reported (14%), all of them on NVE
(19% vs 0%, p=0.03) and one dehiscence of prosthetic heart valve was reported. A new valve
regurgitation was found in 47.2% of IE patients, most commonly on NVE compared to PVE (65.4% vs
15.0%, p<0.001). Native valve insufficiencies were often severe, with a median of severity score of 4/4
[2-4]. In multivariate analysis, the presence of a major echocardiographic Duke criterion, prosthetic
heart valve, vascular/immunological phenomena and a higher LVEDV (left ventricular end-diastolic
volume) were independently associated with a final diagnosis of IE.
Conclusion
A final diagnosis of IE was found in one in four patients referred for suspected IE to our tertiary center.
The disease required surgical treatment on more than 1/3 patients and it was associated with a 13%
in-hospital mortality. The diagnostic accuracy of TTE and TEE in Lausanne was comparable to
previous reports, with a high diagnostic accuracy in NVE but a remarkably low sensitivity for
suspected PVE. Our results therefore confirm the importance of echocardiography as a first-line
diagnostic imaging modality and the emerging role of new advanced imaging techniques including 18FFDG
PET-CT in PVE. Predictors of IE in multivariate analysis were major echocardiographic Duke
criterion, prosthetic heart valve carriers, vascular/immunological phenomena and a higher LVEDV.
These results are in line with the use of the Duke criteria for clinical diagnosis of IE, as recommended
by the current guidelines.