Outcomes following heart valve surgery in patients with infective endocarditis and preoperative septic cerebral embolism: insights from the CAMPAIGN study group

Abstract

OBJECTIVES

This study aimed to analyse the impact of preoperative septic cerebral embolism on early and late postoperative outcomes in patients with infective endocarditis undergoing valve surgery.

METHODS

Retrospective multicentric study based on the Clinical Multicentric Project for Analysis of Infective Endocarditis in Germany (CAMPAIGN) registry comprising patients with infective endocarditis who underwent valve surgery between 1994 and 2018 at 6 German centres. Patients were divided into 2 groups for statistical comparison according to the presence or absence of preoperative septic cerebral embolism. Propensity score matching was performed for adjusted comparisons of postoperative outcomes. Primary outcomes were 30-day mortality and estimated 5-year survival.

RESULTS

A total of 4917 patients were included in the analysis, 3909 (79.5%) patients without and 1008 (20.5%) patients with preoperative septic cerebral embolism. Patients with preoperative septic cerebral embolism had more baseline comorbidities. Mitral valve endocarditis (44.1% vs 33.0% P < 0.001), large vegetations >10 mm (43.1% vs 30.0%, P < 0.001), and Staphylococcus species infection (42.3% vs 21.3%, P < 0.001) were more frequent in the cerebral embolism group. Among patients with preoperative cerebral embolism, 286 (28.4%) patients had no stroke signs (silent stroke). After matching (1008 matched pairs), there was no statistically significant difference in 30-day mortality (20.1% vs 22.8%; P = 0.14) and 5-year survival (47.8% vs 49.1%; stratified log-rank P = 0.77) in patients with and without preoperative cerebral embolism, respectively.

CONCLUSIONS

Preoperative septic cerebral embolism in patients with infective endocarditis requiring valve surgery does not negatively affect early or late mortality; therefore, it should not play a major role in deciding if surgery is to be performed.

Open in new tabDownload slide

INTRODUCTION

Infective endocarditis (IE) is a highly morbid and increasingly common disease, with reported mortality rates of around 20% [1–3]. One of the most feared complications is septembolic stroke, which is observed in up to 50% of IE patients and has been classically associated with an increased risk of death [2, 4–6]. Despite its high prevalence, the impact of preoperative septic cerebral embolism on postoperative outcomes and long-term survival in IE patients requiring valve surgery has been incompletely investigated. This has led to numerous challenges concerning the optimal management of these patients, since the presence of stroke raises the question of when and if to operate. While some argue that the need for surgery is imperative, others are concerned about the potential risk of haemorrhagic transformation due to heparinization during cardiopulmonary bypass [1, 3, 4, 7, 8]. Recent guidelines, however, have suggested that ischaemic stroke should not be a reason to delay or refuse surgery for IE patients, unless the neurologic prognosis is futile [2].

This study aims to analyse the impact of preoperative septic cerebral embolism on the early and long-term outcomes of patients with IE undergoing heart valve surgery based on data from the Clinical Multicentric Project for Analysis of Infective Endocarditis in Germany (CAMPAIGN) Study Group. We hypothesize that preoperative septic cerebral embolism does not have a negative impact on early postoperative morbidity and mortality, as well as on long-term survival.

METHODS

Ethical statement

The study was approved by the ethics committee of the University of Cologne (approval number 17-407) and the individual local ethics committee of each participant centre. Due to the retrospective nature of the study, individual patient informed consent was waived.

Study design and patient population

Retrospective study based on the Clinical Multicentre Project for Analysis of Infective Endocarditis in Germany (CAMPAIGN) registry, including a total of 4917 patients with IE who underwent heart valve surgery between 1994 and 2018 in 6 German centres. There were no formal exclusion criteria. This study of the CAMPAIGN registry analysed the early and long-term outcomes of patients with IE presenting with preoperative septic cerebral embolism. Patients with IE and preoperative septic cerebral embolism were compared against patients with IE and no preoperative cerebral embolism. The primary outcomes were 30-day mortality and estimated 5-year survival. Preoperative septic cerebral embolism was defined as the preoperative occurrence of a silent stroke (i.e. lack of lateralizing neurologic signs) with evidence on cerebral computed tomography (CT) of septic embolism or a clinically manifest stroke in patients with IE. All patients independent of neurological symptoms underwent preoperative whole-body CT examination as part of the standard of care to search for possible infection sources and septic embolisms.

Data collection and follow-up

Each participating centre used the International Classification of Diseases, 10th Revision with German Modification (ICD-10-GM) codes and operation and procedure codes (OPS codes) to identify patients who underwent surgery for IE (Supplementary Material, Table S1). Moreover, administrative and clinical data from the patients’ medical records at every participating institution were used for data collection. The information collected included demographic data and baseline comorbidities, preoperative IE-related clinical manifestations according to the modified Duke criteria (echocardiographic and microbiologic data) [9], intraoperative details, and in-hospital relevant postoperative outcomes and complications. Follow-up was conducted by phone communication with patients, close family members, and/or treating general practitioners/cardiologists in the outpatient setting. The follow-up was closed in the 1st trimester of 2022.

Statistical analysis

Categorical variables are expressed as frequencies and percentages. Continuous variables are expressed as mean and standard deviation for normally distributed variables and median and interquartile range (IQR) for non-normally distributed variables. Given the retrospective nature of the CAMPAIGN registry, there was a minimal amount of data missing at random. We used single imputation to replace missing values. Missing continuous values were replaced with the mean value in normally distributed variables and with the median value in non-normally distributed variables. Missing categorical values were replaced with the mode [10]. Normalcy of distribution was assessed with the Kolmogorov–Smirnov test. Comparisons were performed using crosstabs, the chi-squared, and Wilcoxon rank-sum tests for categorical and continuous variables, respectively. Multivariable unadjusted log regression models were performed to determine early predictors of stroke and in-hospital mortality, with results reported as odds ratios with 95% confidence interval (CI). Similarly, an unadjusted multivariable Cox proportional hazards regression model was performed to determine long-term predictors of long-term mortality. The results of the Cox regression model are reported as hazard ratios with 95% CI. Perioperative variables with an univariable value of P < 0.2 or those judged to be clinically important were submitted to the regression models by backward stepwise selection. Given the statistically significant difference in some clinically relevant preoperative baseline comorbidities and to perform an unbiased analysis of postoperative outcomes, a sub-analysis of the early and late postoperative outcomes after 1:1 propensity score matching was performed. Propensity scores were estimated using logistic regression. We implemented nearest-neighbour matching and confirmed balance by assessing standardized mean differences in covariates between both comparison groups before and after matching. Kaplan–Meier survival curves were computed to analyse the times to event, and the log-rank test was implemented to compare curves between groups during all available follow-up. Matched groups were compared using the stratified log-rank test. Standardized mean differences (SMDs) were calculated for additional comparison between variables. P-values <0.05 were considered statistically significant for all tests. All statistical analyses were performed using R (Version 2024.04.1 + 748).

RESULTS

A total of 4917 patients underwent cardiac surgery for IE at 6 German centres during the study period. Amongst them, 3909 (79.5%) patients had no preoperative cerebral embolism, and 1008 (20.5%) had a preoperative septic cerebral embolism.

Demographic and preoperative characteristics

Patients’ demographic characteristics are shown in Table 1. The cerebral embolism group had a higher median EuroSCORE II than the non-cerebral embolism group [11% (IQR 5–24%) vs 10% (IQR 5–17%); P = 0.007]. There was a significant difference in the prevalence of baseline cardiovascular disease, with a higher disease burden in the cerebral embolism group. The need for preoperative mechanical ventilation was also significantly higher in the cerebral embolism group (18.1% vs 7.2%, P < 0.001). Among patients with preoperative cerebral embolism, a total of 722 (71.6%) had a clinically manifest stroke, while 286 (28.4%) had a silent stroke. Supplementary Material, Table S2 shows the preoperative characteristics of patients with silent and manifest stroke. Although the baseline profile was very similar overall, the manifest stroke group presented with a higher prevalence of preoperative dialysis (10.1% vs 5.6%, P = 0.23) and mechanical ventilation (19.9% vs 13.3%, P = 0.13).

IE-related details are summarized in Table 2. Patients in the cerebral embolism group had a higher prevalence of mitral valve IE (44.1% vs 33.0% SMD = 0.23, 95% CI 0.16–0.3, P < 0.001), more often presented with vegetations (87.8% vs 57.9%, SMD = 0.71, 95% CI 0.54–0.78, P < 0.001) and showed a higher prevalence of large vegetations (i.e. >10 mm) (43.1% vs 30.0%, P < 0.001). Evidence of Staphylococcus spp. as the causative microorganism was significantly higher in the cerebral embolism group (42.3% vs 21.3%, SMD = 0.27, 95% CI 0.21–0.34, P < 0.001). The overall incidence of non-cerebral systemic emboli was also higher in the cerebral embolism group (Table 2).

Intraoperative data

Biological aortic valve replacement was the most commonly performed procedure in both groups (n = 1721, 44.0% in the non-cerebral embolism group and n = 504, 50.0% in the cerebral embolism group), followed by biological mitral valve replacement (n = 902, 23.1% in the non-cerebral embolism group and n = 392, 38.9% in the cerebral embolism group). Concomitant coronary artery bypass was more commonly required in the cerebral embolism group (16.3% vs 11.5%, P < 0.001). Further intraoperative data can be found in Table 3.

Postoperative outcomes

In order to account for differences in baseline variables, we performed propensity score analysis matching for those characteristics that were shown to be significantly different between both groups. A total of 1008 patients were matched in a 1:1 ratio. Balance was achieved for all covariates, as evidenced by standardized mean differences below 0.1 (Supplementary Material, Fig. S1 and Supplementary Material, Table S3).

Early postoperative outcomes

The early postoperative outcomes for the matched and unmatched cohorts are presented in Table 4. Before matching, the median mechanical ventilation time was longer in the cerebral embolism group [25 h (IQR 10–162 h) vs 15 h (IQR 8–95 h), P < 0.001] and the median intensive care unit (ICU) length of stay was significantly longer amongst patients from the cerebral embolism group [4 (IQR 1–10) vs 3 (IQR 1–5) days, P < 0.001]. Moreover, patients from the cerebral embolism group more frequently presented with postoperative complications, such as new-onset postoperative stroke (24.9% vs 12.0%, P < 0.001).

The unadjusted 30-day mortality was 20.1% for patients with cerebral embolism and 16.9% for patients without cerebral embolism (P = 0.01) (odds ratio 1.13; 95% CI 0.97–1.33; P = 0.11). After propensity score matching, there was no statistically significant difference in the 30-day mortality (22.8% vs 20.1%; SMD = 0.07, 95% CI 0.02–0.15, P = 0.14/odds ratio 1.00, 95% CI 0.84–1.2, P = 0.96) and median ICU length of stay [3 (IQR 1–9) vs 4 (IQR 1–9) days; SMD = 0.07, 95% CI 0.02–0.15, P = 0.11] of the non-cerebral and cerebral embolism groups, respectively. However, the incidence of new-onset postoperative stroke remained significantly higher in the cerebral embolism group (24.9% vs 15.4%, SMD = 0.24, 95% CI 0.15–0.33, P < 0.001). Predictors of preoperative cerebral embolism and 30-day mortality derived from the log regression model are shown in Tables 5 and 6.

Late-postoperative outcomes and follow-up

The follow-up completeness rate was 86.3%, with a median follow-up of 763 days. The coefficient for the estimated effect of preoperative cerebral embolism on survival was –46.5 (95% CI –159 to 65.8, P = 0.41) based on g-computation. Figure 1A depicts the unadjusted Kaplan–Meier estimated survival, demonstrating a significantly higher long-term survival probability during all available follow-up amongst patients without preoperative cerebral embolism, as demonstrated by the log-rank test (P < 0.001). The estimated unadjusted survival at 5 years was 64.6% in the non-cerebral embolism group versus 47.8% in the cerebral embolism group (hazard ratio 1.23, 95% CI 1.10–1.38, P < 0.001). After propensity score matching, there was no statistically significant difference in 5-year survival among patients with and without cerebral embolism: 49.1% vs 47.8%, respectively (stratified log-rank P = 0.77; hazard ratio 1.02, 95% CI 0.89–1.16, P-value = 0.80) (Fig. 1B). In the unmatched cohort, there was no significant difference in the estimated survival of patients according to the clinical type of cerebral embolism (i.e. 5-year survival was 44.5% in patients with a silent stroke versus 49.1% in patients with a manifest stroke; log-rank P = 0.59). When plotted against the matched cohort of patients without cerebral embolism, the survival remained similar (log-rank P = 0.84) (Supplementary Material, Fig. 2). Predictors of long-term mortality derived from the Cox proportional hazards model are shown in Table 7.

DISCUSSION

The current analysis of data from the multicentric CAMPAIGN registry assesses the impact of preoperative septic cerebral embolism on the early and long-term outcomes of 4917 patients with IE who underwent heart valve surgery over a 24-year period in 6 German cardiac surgery centres. The main findings of this study are:

1. Preoperative septic cerebral embolism is a frequent complication of IE, observed in about one-fourth of patients undergoing surgery. One-third of patients with cerebral embolism have a ‘silent’ stroke.

2. IE patients presenting with preoperative cerebral embolism and those patients presenting with manifest stroke have a higher prevalence of baseline comorbidities and an increased cardiovascular disease burden.

3. Mitral valve IE, large vegetations (>10 mm) and Staphylococcus spp. infection are more frequent in patients with preoperative cerebral embolism.

4. After matching for baseline characteristics, the 30-day mortality and 5-year survival rates are not negatively affected by preoperative cerebral embolism.

About 1 in 4 patients from the CAMPAIGN registry presented with preoperative septic cerebral embolism, which is in line with previous studies reporting evidence of acute cerebral septic emboli in ∼20–50% of patients with active IE [2, 11]. Similarly, other reports observed that almost one-third of patients with IE diagnosed with stroke had a so-called ‘silent’ stroke, characterized by the absence of lateralizing symptoms in the presence of an embolic ischaemic stroke on CT [7, 12, 13]. Patients with silent or manifest stroke had a similar baseline profile, and the presence of clinically manifest stroke did not have a significant impact on long-term survival. However, we observed major differences in their baseline profile when comparing patients based on the presence or absence of preoperative septic cerebral embolism, regardless of their neurological symptoms. Patients from the cerebral embolism group had a considerably higher baseline cardiovascular disease burden (i.e. higher incidence peripheral artery disease and hypertension) and more complex disease (i.e. left-sided IE, larger vegetation sizes, higher incidence of Staphylococcus spp. infection, need for mechanical ventilation). Some of these factors, for example infection with Staphylococcus spp. and vegetation size >10 mm, have been repeatedly described in the literature as potential predictors of stroke [2, 5, 14, 15]. Other factors that were also more frequently observed in patients from the cerebral embolism group such as smoking, hypertension and peripheral artery disease, have been less discussed in the context of stroke and IE, but are known risk factors for generalized atherosclerosis. Our multivariable logistic regression model demonstrated that mitral valve IE, need for mechanical ventilation, arterial hypertension, large vegetation size and Staphylococcus infection strongly correlated with cerebral embolism in the CAMPAIGN cohort.

The presence of stroke has been previously described as an additional predictor of poor outcomes in patients with IE, with some studies reporting almost twice the risk of postoperative death [2, 4–6]. Indeed, patients from the preoperative cerebral embolism group presented with lower short- and long-term unadjusted survival in the current study. However, given the marked discrepancies in baseline characteristics between groups, with a higher disease burden in the cerebral embolism group, we performed a propensity score analysis matching groups for baseline characteristics to assess the independent impact of cerebral embolism on survival. Based on the P-value for the estimated effect of cerebral embolism and the result of the stratified log-rank test comparing the resulting Kaplan–Meier curves of the matched cohorts, there was no evidence to conclude that the occurrence of septic cerebral embolism itself has a significant impact on either early or late mortality. We believe that the observed difference in the unmatched cohort is related to the poorer baseline health status from the cerebral embolism group, which could explain the reduced survival. This conclusion is supported by the results of the Cox- and Log regression models, which revealed several of these comorbidities as short and long-term predictors of mortality.

The current cohort came from a large database with patients from various centres and unfortunately, there was no retrospective information regarding the time interval between cerebral embolism and surgery. However, given that the data came from a real-world registry, it is accurate to assume that it is a mixed pool of patients who underwent early and delayed surgery following cerebral embolism. The higher morbidity in the early postoperative period and the prolonged ICU times in the cerebral embolism group are unlikely explained by the timing of surgery, but rather by the high-risk surgical profile of these patients, as suggested by the results from the matched cohort. It is, however, noteworthy that after matching, the incidence of postoperative bleeding and the need for tracheotomy remained slightly higher in the cerebral embolism group. The cause of the increased bleeding rates remains unclear. The increased need for tracheotomy is related to the naturally expected more delayed transition to spontaneous ventilation in patients with neurological deficits secondary to stroke. The risk of repeated embolism, uncontrolled infection, multiorgan dysfunction and septic shock can be mitigated through surgical excision of the infected cardiac tissue. Hence, once there is a surgical indication in patients with IE, prolonged conservative therapy with antibiotics will unlikely improve the clinical outcomes or negate the need for surgery.

Since septic embolic stroke did not have a negative impact on early mortality and long-term survival in our cohort of patients with IE, we firmly believe that preoperative cerebral embolism alone should not be a reason for not performing or delaying surgery. The 2023 ESC/EACTS guidelines for the management of patients with IE advocate for evaluation of all IE patients by an Endocarditis Team as a IB recommendation [2]. Patients with preoperative cerebral embolism are a particularly complex and high-risk group, and transfer to a Heart Valve Centre is recommended for all such patients [2]. A multidisciplinary Endocarditis Team consisting of a wider variety of specialties, including cardiac surgery and neurology, should be involved in the decision-making process of such patients at Heart Valve Centres, but our data confirm that preoperative cerebral embolism should not be the only decisive factor in the management of these patients [16]. Indeed, the 2023 IE guidelines clearly state that in patients in whom a surgical indication exists, such as those with heart failure, uncontrolled infection or high risk of further embolization, the presence of preoperative cerebral embolism should not lead to a delay in surgical management, provided that neurologic prognosis is non-futile [2]. Patients with complex IE and paravalvular involvement [17, 18] or IE of transcatheter or conventional prosthetic valves [19, 20] are particularly high-risk, and delay of surgical therapy is not recommended once an indication for surgery has been confirmed. However, surgery should be delayed in those patients presenting with significant intracerebral haemorrhage [2].

Limitations

This study has several limitations, mainly due to its retrospective nature, which makes it inherently subject to biases like selection bias. Furthermore, the variability of the follow-up methods, in addition to a study period spanning over 24 years, could lead to a lack of homogeneity of the collected data and discrepancies in outcomes after time stratification, which, when unaccounted for, could translate into information bias. However, time stratification did not significantly alter our study outcomes, given that diagnostic and treatment of IE have not drastically changed in the last decades. Moreover, our registry did not contain information on the time interval between the occurrence of preoperative cerebral embolism and surgery. However, our study is an analysis of an ‘all comers’ population that comprises patients who were operated on early and late after preoperative cerebral embolism, and it therefore reflects real-life clinical practice. Despite CT being the standard of practice for the acute diagnosis of stroke in routine clinical practice, CT examination has limited sensitivity for the assessment of early cerebral infarct signs, which might lead to under-detection of stroke events. Furthermore, there was a lack of information regarding the specific tomographic characteristics and the extent of the preoperative stroke. Detailed information on the neurological manifestations and neurological functional status of the patients after the cerebral embolic event, before discharge, and during follow-up is also missing, precluding further analyses of these outcomes.

CONCLUSIONS

In conclusion, patients with IE and preoperative septic cerebral embolism have similar early mortality and 5-year survival compared to patients without preoperative septic cerebral embolism. The patient’s general condition, specific characteristics, and clinical presentation need to be thoroughly evaluated for decision-making. The occurrence of preoperative cerebral embolism should not be the main factor influencing the decision to delay or not perform surgery.

SUPPLEMENTARY MATERIAL

Supplementary material is available at EJCTS online.

FUNDING

None declared.

Conflict of interest: Michael A. Borger discloses that his hospital receives speakers’ honoraria and/or consulting fees on his behalf from Edwards Lifesciences, Medtronic, Abbott, and CryoLife. The remaining authors have no conflicts of interest or financial relationships with the industry to disclose.

DATA AVAILABILITY

The data underlying this article will be shared on reasonable request to the corresponding author.

Author contributions

Mateo Marin-Cuartas: Conceptualization; Data curation; Formal analysis; Investigation; Methodology; Project administration; Validation; Visualization; Writing—original draft; Writing—review & editing. Manuela de la Cuesta: Formal analysis; Methodology; Validation; Writing—original draft; Writing—review & editing. Carolyn Weber: Writing—review & editing. Elisabeth Krinke: Writing—review & editing. Artur Lichtenberg: Writing—review & editing. Asen Petrov: Writing—review & editing. Christian Hagl: Writing—review & editing. Hug Aubin: Writing—review & editing. Klaus Matschke: Writing—review & editing. Mahmoud Diab: Writing—review & editing. Maximilian Luehr: Writing—review & editing. Payam Akhyari: Writing—review & editing. Philipp Schnackenburg: Writing—review & editing. Sems-Malte Tugtekin: Writing—review & editing. Shekhar Saha: Writing—review & editing. Torsten Doenst: Writing—review & editing. Thorsten Wahlers: Writing—review & editing. Michael A. Borger: Supervision; Writing—review & editing. Martin Misfeld: Supervision; Writing—review & editing.

Reviewer information

European Journal of Cardio-Thoracic Surgery thanks Catalin Constantin Badiu, Samuel Heuts and Cristian Dinges for their contribution to the peer review process of this article.

Presented at the 37th Annual Meeting of the European Association for Cardio-Thoracic Surgery, Vienna, Austria, October 2023.

REFERENCES

ABBREVIATIONS

ABBREVIATIONS
 
• CI

  Confidence interval

 
• CT

  Computed tomography

 
• ICD-10-GM

  International Classification of Diseases, 10th Revision with German Modification

 
• ICU

  intensive care unit

 
• IE

  Infective endocarditis

 
• IQR

  Interquartile range

 
• SMD

  Standardized mean difference

Author notes