Common indications for cardiac surgery in patients with infective endocarditis (IE) include refractory heart failure (HF), disseminating infection with periannular extension, multi-resistant microorganisms, recurrent embolic events and the presence of prosthetic material.1 Among the aims of surgery are excision of infected tissue, restoring cardiac integrity and valve function, and removal of potential sources of emboli.2 Irrespective of the surgical approach and the procedures used, such as mitral valve repair or homograft aortic root replacement, long-term results of emergent and urgent surgery are worse than those from elective surgery: 10-year survival ranges from 40% to 60%.3

This study had two aims: to characterize patients undergoing cardiac surgery for IE in sociodemographic and clinical terms, including the type of valve and microbiological agents involved, type of surgery and surgical indications, complications and mortality; and to identify predictors of in-hospital mortality.

This study focuses on two aspects of this patient population which enable comparisons with results from similar studies: clinical characterization of IE patients undergoing surgery, and predictors of in-hospital mortality. With regard to clinical and demographic characteristics, male patients were more affected by IE (72.4%) than females, the most frequent microorganisms isolated were Staphylococcus aureus followed by Streptococcus spp., and the native aortic valve was most often involved. Surgery was most often urgent (within seven days)5 and the main indication was refractory HF, followed by large vegetations. The most common surgical complications were uremia requiring hemodialysis, acute kidney injury and third-degree AV block. Overall in-hospital mortality was 13.1% and septic and cardiogenic shock were the main causes of death. The main predictors of higher in-hospital mortality were the need for emergent surgery,5 lower preoperative LVEF, AF, pre- and postoperative sepsis or septic shock, cardiogenic shock, cardiac tamponade, and the need for renal replacement therapy in the postoperative period.

In this study, in-hospital mortality was 13.1%, which is similar to results from other retrospective studies such as Moreira et al.,8 with 15% mortality in 133 patients who underwent cardiac surgery for IE, and Farag et al.1 (18.3% in 360 operated patients), but higher than in Ferreira et al. (3.7% in 54 patients who underwent surgery for IE).9 Predictors of in-hospital mortality in this study were different from those described by Moreira et al.8 (chronic obstructive pulmonary disease, cerebral embolism, IE caused by Staphylococcus spp. and non-HACEK Gram-negative bacilli), by Ferreira et al.9 (operated and non-operated patients with IE: Charlson index ≥5, EuroSCORE >6, use of immunosuppressants, severe sepsis and/or septic shock, and inappropriate antibiotic therapy), and by Farag et al.1 (advanced age, preoperative diabetes, higher NYHA class, kidney failure and liver failure, longer operative, cardiopulmonary bypass (CBP) and aortic cross-clamp times, greater volume of red blood cells transfused, and fresh plasma and platelet transfusion). Many of the variables in these studies1,8,9 were not assessed in ours. Differences in baseline characteristics of the IE patients included in these studies may explain why they found predictors of mortality that were different from those in our study. For example, in Moreira et al.,8 diabetes, chronic kidney disease (CKD) and Staphylococcus spp. as the etiological agent were less frequent, while a history of endocarditis and the presence of vegetations >10 mm were more common. In the studies by Farag et al.,1 Moreira et al.,8 Ferreira et al.,9 mean patient ages were lower.

Certain comorbidities may have a significant impact on mortality. In our study, only eight patients had a history of stroke (5.5%) and all survived. CKD (creatinine clearance <60 ml/min/1.73 m2) was diagnosed in 13.8%; degree of renal dysfunction had no significant effect on survival, but mortality was significantly higher in patients requiring postoperative dialysis (78.9% vs. 16.7%; p<0.001). The proportion of patients with recent stroke and/or CKD was lower in our study than in Farag et al.1 Although fewer patients in our study had a history of stroke, and thus practical conclusions cannot be drawn, in Misfeld et al.10 patients with IE and cerebral embolism had a five-year survival of 46%, which was lower than in those without cerebral embolism. Among patients discharged from hospital, overall survival is over 60% at 10 years. Farag et al.1 found a mean survival of 69.4% at one year, 63.3% at five years and 63.3% at 10 years following cardiac surgery for IE.

The present study only considered patients who underwent cardiac surgery for IE and not patients admitted to the same center who were only treated medically. According to Cahill et al.,2 50–60% of patients with IE undergo cardiac surgery, while Moreira et al.8 observed that in-hospital mortality in patients with IE who underwent cardiac surgery was significantly lower than in those who did not (15.5% vs. 32.6%; p=0.028), as also observed by Ferreira et al.9 Our study did not make this comparison, but it would have been useful to do so in order to enable conclusions to be drawn on various parameters, including surgical indications and the most appropriate timing for surgery.

Thirty-day mortality following surgery for IE ranges from 15% to 20%,1,11–13 which is similar to the results from the present study.

HF was the main surgical indication for most of the patients included in the study (57.9%), as also observed by Farag et al.1 (around half their patients were in NYHA class IV). The main mechanisms underlying HF were severe valve regurgitation and valve dysfunction. Uncontrolled infection (large vegetations and abscesses) and the prevention of cardioembolism were other predictors of cardiac surgery. The ESC5 and Cahill et al.2 identify these three factors as the surgical indications with potential benefit for IE patients, which need to be identified early to improve surgical outcomes and thereby reduce early and long-term mortality. All the patients in our study cohort underwent cardiac surgery, but data on potential candidates who were not operated were not analyzed. According to Chu et al.,13 24% of candidates with guideline-recommended indications for surgery are not actually operated. Among the reasons given are poor prognosis irrespective of treatment (34%), hemodynamic instability (20%), death before surgery (23%), stroke (23%), sepsis (21%), and refusal for surgery (26%).2 In our study, surgery was most often urgent (within a few days)5 (74.5%) (76.2% in survivors and 63.2% in non-survivors (p=0.21), and emergent surgery was performed in 17.5% of patients in survivors and 36.8% in non-survivors (p=0.065) (Figure 5). Moreira et al.8 concluded that early cardiac surgery in IE (within seven days) was not associated with higher mortality compared to later surgery.

The profile of the microbial agents found in this study and their relative proportions were similar to those found in Farag et al.1 – predominantly Staphylococcus spp. (45%) (MRSA 25%, MSSA 15% and S. epidermidis 5%). Not all the tissues excised during surgery underwent histopathological or microbiological analysis (which is now the cornerstone of IE diagnosis),5,14 nor was the antibiotic therapy used in the postoperative phase characterized. These data are important to help evaluate some predictors of mortality.

As pointed out by Farag et al.,1 the type of microorganism isolated has changed, with Staphylococcus spp. now predominating as etiological agents, which due to their more aggressive and invasive nature may lead to the formation of large vegetations and abscesses. These alone are an indication for surgery (observed in 29.7% of cases in the present study). IE caused by Staphylococcus has repeatedly been associated with higher mortality,15,16 although this was not observed in our study.

Many other variables may be associated with higher mortality in cardiac surgery for IE patients. Several authors17,18 have demonstrated a relationship between the presence of vegetations >10 mm, large abscesses, false aneurysms and fistulas,11,19,20 and adverse events and prognosis, but in our study mortality was not significantly higher in the presence of these indicators of uncontrolled infection (although our small sample size should be taken into account).

The simultaneous presence of severe valve regurgitation and cardiogenic shock is associated with higher 30-day mortality. According to the ESC guidelines,5 surgery must be performed on an emergency basis,21 irrespective of the status of infection, when patients are in persistent pulmonary edema or cardiogenic shock despite medical therapy.22 As seen in the present study, mortality is higher in such patients. Surgery should be performed urgently when HF is less severe or there is severe aortic or mitral valve regurgitation with large vegetations, even in the absence of HF.23

As seen in other studies,1,8 native valves were most frequently affected, followed by biological and mechanical valves; however, no significant differences were recorded in in-hospital mortality according to the type of valve. This contrasts with the higher mortality found by other authors with prosthetic valves.1,24

The main cardiac complications before surgery for IE in this study were similar in type to those observed in other studies1,5,8 (in most cases urgent or emergent, and not uncommonly life-saving), but differ in their frequency, especially for severe valve regurgitation, refractory HF (15% in this study), large vegetations and valve dysfunction. These differences may be part of the reason for the different predictors of postoperative mortality in our study compared to the literature, as discussed above. Extracardiac complications, the most frequent of which were systemic embolism, respiratory failure, sepsis and renal dysfunction, in themselves present a range of preoperative risks and are observed in varying proportions in different studies.1,8 They influence the approach toward anesthesia, transfusion (and its effect on coagulation parameters), medical therapy and mechanical circulatory support, and affect decisions on what types of surgery are possible or desirable in terms of operative, cardioplegia, CBP and aortic cross-clamp times,1 type of valve or homograft to be implanted, extent of debridement or resection of intracardiac tissues, including paravalvular tissue, and possible involvement of conduction tissue and/or the valvular annulus, use of Dacron patches, and level of hypothermia.1 The severity of these complications varies so much, as do their impact on postoperative mortality, that it is difficult to compare predictors reported in studies published over the last 20 years. Some of these factors, according to Farag et al.,1 are the most effective predictors of mortality: operating, cardioplegia, CBP and aortic cross-clamp times, and massive transfusions (and their impact on coagulation parameters).

Biological valves were selected for most patients (62.1%); the reasons for this choice (such as age, need for oral anticoagulation, bleeding risk, and expected treatment adherence) were not stated on the medical records. In Lee et al.’s study,25 biological valves were used in the majority of patients (75.8%, n=71) requiring mitral valve replacement, and mechanical valves were used in 24.3%.

In our study, AF, sepsis and cardiac tamponade were the most common postoperative complications in patients who died during hospital stay. According to David et al., 26 the most common postoperative complications following surgery for IE are severe coagulopathy, repeat chest surgery for bleeding or cardiac tamponade (in our study associated with 15.8% mortality in the non-survivor group compared to 1.6% in the survivor group; p=0.01), aute kidney injury or uremia requiring hemodialysis (in our study, the most common postoperative complication; 24.8% of patients required dialysis with markedly higher associated in-hospital mortality: 78.9% vs. 16.7%; p<0.001), HF (again corroborated by our study, especially with regard to concomitant severe valve regurgitation and cardiogenic shock (35.7% vs. 5.7%; p=0.005), pneumonia, and third-degree AV block following radical resection of the aortic root due to an abscess and requiring pacemaker implantation.26

In our study, new-onset AF was less common (8.3%) in the postoperative period than reported in the literature (15–45%).27 Postoperative AF is associated with longer hospital stays and higher rates of complications and mortality28 (as found in our analysis).

This study only assessed in-hospital mortality (13.1%), for which septic shock and cardiogenic shock were the main causes. This figure is slightly lower than that in Farag et al.1 (18.3% at 30 days). In the study by Kang et al.,23 patients with IE who underwent early cardiac surgery were significantly less likely to have the composite endpoint of all-cause mortality, embolic events or recurrence of IE at six months. Farag et al.1 reported postoperative IE survival of 63.3% at five years, 63.3% at 10 years and 48.3% at 20 years.

This study and others show that, in the context of IE, surgery often plays a crucial and often life-saving role,14 such as in cases of rupture of the chordae tendineae, severe valve regurgitation, biological or mechanical valve dysfunction, refractory HF, sepsis unresponsive to antibiotics, large intracardiac vegetations and abscesses, and risk of cardioembolism.

In this study, several topics were not studied, such as the length and characteristics of patients’ hospital stay, including length of stay between the initial diagnosis and surgery and between surgery and discharge or in-hospital death, the duration of other treatments in the pre- and postoperative periods, and potential indicators of treatment quality,14 as well as complications which may have prolonged hospital stay and influenced in-hospital and medium- and long-term mortality.14

Due to the retrospective nature of our study, we were unable to investigate several items on clinical records which are essential to calculate the scores used to estimate mortality risk following cardiac surgery in patients with IE, such as the Association pour l’Etude et la Prévention de l’Endocardite Infectieuse (AEPEI), modified AEPEI and PALSUSE scores, and to analyze predictors of in-hospital mortality, particularly the parameters used to calculate EuroSCORE II, such as those related to whether the patient is considered to be in a critical state, which is also included in the AEPEI and PALSUSE scores. We were therefore unable to validate this excellent risk prediction tool in the IE patients included in our analysis. In the future, a prospective study could answer many of the questions raised by the results of this study.

This study has several limitations. It was a retrospective analysis (and therefore potentially important information was missing or incomplete in clinical records) on a relatively small sample over 11 years, a period during which changes took place in surgical approaches and criteria, techniques and prosthetic material, diagnostic methods, and intensive postoperative care that will undoubtedly have affected the quality of care and hence in-hospital mortality. Over this period patients also presented with increasingly frequent and severe comorbidities. Certain important data related to the surgical procedure itself are also missing, including operative, cardioplegia, CBP and aortic cross-clamp times; extent of debridement or resection of intracardiac tissues, including paravalvular tissue, and possible involvement of conduction tissue and/or the valvular annulus; use of Dacron patches; and level of hypothermia. In addition, we did not analyze the duration of pre- or postoperative antibiotic therapy, length of stay in other hospitals, or time between admission of a patient with indication for surgery to another hospital and the operation.

In our study, 13% of patients with IE died in the postoperative period. There is a need for better indicators to enable early identification of surgical candidates for IE; implementation of a heart team composed of a cardiologist (clinical or experienced in transthoracic and transesophageal echocardiography), a cardiothoracic surgeon, an internist/intensivist, an infectious disease specialist, a neurologist, a primary care physician and an imaging specialist; and better surgical strategies, including more rapid intervention, more specific postoperative care and optimal antibiotic therapy, to reduce in-hospital and medium- and long-term mortality.

In practical terms, early cardiac surgery (within seven days) should be considered by the heart team when Staphylococcus spp. are isolated in blood cultures, or in the presence of refractory HF or large vegetations (>10 mm), septic and/or cardiogenic shock, or LVEF<50%.

The authors have no conflicts of interest to declare.