This is a single-center retrospective observational study conducted in a tertiary referral hospital, the First Affiliated Hospital of Sun Yat-sen University. From January 1st of 2008 to September 30th of 2017, a total of 480 consecutive patients with clinical suspicion of IE were screened. We excluded patients who were younger than 18 years old (n=32), without definite IE (n=34) according to the modified Duke Criteria were included in the study , and without comprehensive echocardiographic evidence of endocarditis (n=32). Patients with evidence of right-sided endocarditis (n=37), cardiac implantable electronic devices (pacemakers and implantable cardioverter-defibrillators) (n=4), both-sided IE (n=11), and congenital heart disease-related IE (n=21) were also excluded, given these subtypes have completely distinct clinical, microbiological, and prognostic characteristics from LSIE. We also excluded 102 patients who didn’t undergo IE surgery. Finally, a total of 207 patients ≥18 years old with a definite LSIE diagnosis undergoing LVS were included in the study (Figure 1). Patients were further divided into two groups depending on the type of definitive interventions they received: LVS group (n=157), and LVS+TA group (n=50). The LVS group conducted left-sided valve surgery alone, while the LVS+ TA group underwent concomitant TA with LVS.
Definition of terms
The estimated glomerular filtration ratio (eGFR) was calculated using the 2009 CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration) creatinine (eGFRcr) equation . The presence of heart failure was determined by symptoms and signs elicited from clinical assessment, laboratory investigations, and radiographic findings at baseline preoperatively [17, 18]. Systemic embolisms were defined as acute onset of clinical symptoms or signs indicating a lodging embolus involving the brain, spleen, kidneys, or lung, consistent with radiographic evidence . Vascular phenomena, such as cutaneous microinfarctions and metastatic abscesses, were not classified under systemic embolisms. Neurological complications included meningoencephalopathies, ischemic complications, cerebral hemorrhage, and intracranial abscess . According to previous studies and guidelines, paravalvular complications were defined as intracardiac abscesses or fistulas based on transthoracic or transesophageal echocardiograms [21, 22]. Postoperative complications included major bleeding, neurological complications, cardiac complications, infection, and thoracic effusion or pneumothorax and poor wound healing. Major bleeding was defined as: 1) intracranial bleeding; 2) overt bleeding resulting in a decrease in hemoglobin ≥20 g/L or requiring blood transfusion; or 3) bleeding into a confined space, such as pericardial cavity or paraspinal space, which indicates severe morbidity. The definition of neurological complications was above-mentioned. Cardiac complications included heart failure, arrhythmia, massive pericardial effusion, and low cardiac output syndrome. The study outcome for this analysis was in-hospital and long-term all-cause mortality. Since the high in-hospital mortality rate, the long-term all-cause mortality was calculated from both index admission and from discharge respectively.
Data collection and measurement of echocardiographic parameters
Baseline data were collected from the electronic medical record system, including the patients’ demographic characteristics, previous history of underlying diseases, clinical presentations, laboratory test results on admission, transthoracic/transesophageal echocardiographic data, complications and microscopic studies of associated pathogens. The clinical severities were assessed by the Pitt bacteremia scores (PBS) ranging from 0 to 14 points with the higher score indicating the severer status .
Transthoracic echocardiograms (TTE) were performed at baseline preoperatively and postoperatively. The median duration of postoperative TTE performed were 8 (inter-quartile range, IQR: 7-13) days post operation. Left ventricular ejection fraction (LVEF) was measured by Simpson’s method. In accordance of the recommendations from American Society of Echocardiography, the severity of tricuspid regurgitation was assessed using color Doppler flow images in TTE and graded qualitatively as mild (1+), moderate (2+), or severe (3+ or 4+) following the jet area-central jets of < 5, 5- 10, or > 10 cm2 respectively . The pulmonary artery systolic pressure was calculated as the addition of estimated right atrial pressure (ranging from 5 to 10 mmHg given the varied size of inferior vena cava) and the systolic right atrial-ventricular pressure gradient (PG/ΔP), which was calculated by the modiﬁed Bernoulli equation: ΔP = 4 × v2 (v: the maximal velocity of the TR jet area) .
All operations were performed under conventional cardiopulmonary bypass, mild or moderate hypothermia through median sternotomy. Mitral or aortic valve surgeries were performed before examining the tricuspid valve. For patients who underwent concomitant TA, either De Vega annuloplasty or ring annuloplasty were used according to the surgeon’s comprehensive evaluations following the latest guidelines’ recommendations at that time [26-28].
Continuous normally distributed variables were described as mean value with standard deviation (SD) or median (IQR) when appropriate. For quantitative variables, the groups were compared by a two-tailed Student’s t-test or Mann-Whitney U-test when necessary. Categorical variables were expressed as number of event and a percentage, and they were examined by the χ2 test or Fisher’s exact test when appropriate. Categorical echocardiographic variables change between pre- and post-operation were analyzed using McNemar's statistical test.
Multivariable logistic regression models were applied to determine whether concomitant TA were associated with in-hospital mortality postoperation after adjusting for confounding factors. Multivariable cox regression models were used to explore the association of concomitant TA with long-term survival adjusted for predefined covariates. The survival time was calculated from death or the last time of follow up to hospital discharge. Since the high rate of in-hospital mortality, the survival time between death or last time of follow up to hospital admission were also discussed.
All hypothesis tests were two-sided, and a P value < 0.05 was considered as statistically significant. Statistical analysis was performed with SPSS software V22.0 (SPSS Inc., Chicago, IL, USA).