Neurologic sequelae are among the most common and devastating complications of infective endocarditis (IE) with an incidence varying between 10% and 75% [ ] and mortality ranging between 20% and 58% [ ]. In particular, mortality is higher in those with neurological complications than in those without [ , ], prompting careful attention to the management of patients with cerebral injuries. The spectrum of neurological sequelae with their approximate proportions includes ischemic stroke (70%) and intracerebral hemorrhage (10%). Less prevalent complications such as subarachnoid hemorrhage, meningoencephalitis, and intracerebral abscess each represent approximately 5% of cases. Currently, surgery is indicated in IE patients with uncontrolled sepsis despite antibiotics (such as those with aortic root abscess, ongoing fever, recurrent embolization events while on antibiotics) and those with severe valve regurgitation causing heart failure [ , ]. Based on these indications, approximately 50% of patients suffering from IE undergo cardiac surgery in combination with antibiotics therapy [ ]. However, in patients whose presentation was complicated by a stroke related to IE, a careful assessment of the risks and benefits is required.
The decision on surgery requires a balance between the urgency of the operation for cardiac indications and the perceived risk of exacerbating the neurological injury with intracerebral hemorrhage that complicates existing hemorrhagic stroke or hemorrhagic conversion of an ischemic stroke, hypotension, or further embolization related to cardiopulmonary bypass. Few large prospective studies have provided guidelines regarding the optimal timing for surgical intervention after stroke in IE. This chapter will cover the evolution of data assessing outcomes of cardiac surgery after a neurological event related to IE and provide current societies guidelines on the subject.
Ischemic stroke and timing of surgery
Looking closely at ischemic stroke, the most common neurological complication after IE, it is worth noting that earlier studies reported higher mortality with surgery performed in the acute phase after presentation. A multicenter retrospective study of 181 Japanese patients with IE who suffered a stroke showed decreased rate of neurological deterioration with delayed surgical intervention: 45.5% if surgery happened within 24 h, 43.8% if performed within a week, 16.7% if performed between 1 and 2 weeks after initial insult, and 2.3% if more than 4 weeks after an ischemic stroke [ ]. The authors concluded that cardiac operations can be safely conducted around 4 weeks after the initial infarction. Garcia-Cabrera et al. also reported that the risk of further cerebral injury was inversely correlated with time of surgical intervention: the risk of intracranial bleeding was 50%, 33%, and 20% if surgery was completed at <2, 2–3, and >3 weeks, respectively [ ]. Similar conclusions were reached by Angstwurm et al. who reported a 20% risk of neurological exacerbation with surgery performed within 3 days of presentation, 20%–50% risk if surgery between 4 and 14 days, and <10% if when surgery was done >14 days, and <1% if surgery was done >4 weeks after initial embolic event [ ].
However, data are conflicting as recent reports of early surgery following IE with stroke show no adverse outcomes, leading to a paradigm shift toward early surgery with special consideration for patients suffering for intracranial hemorrhage (ICH). In 2015, Sorabella et al. determined that in a cohort of patients who underwent early surgery defined at ≤14 days, there was no difference in rate of new postoperative stroke between patients with and without prior embolic stroke (9.3% vs. 7.1%, P = .57) [ ].
Surgical mortality in patients with IE has greatly improved, and regardless of neurologic sequelae, recent data indicate that surgery in the acute phase confers a survival benefit. In a meta-analysis of 21 studies, Yanagawa et al. showed a survival advantage with early surgery (performed <7 days) for both unmatched (odds ratio [OR] 0.61; 95% confidence interval [CI] 0.50–0.74; P < .001) and propensity-matched cohorts (OR 0.41; 95% CI 0.31–0.54; P < .001). Similarly, in a prospective randomized trial, Kang et al. also found reduced all-cause mortality with surgery performed less than 48 h (hazard ratio 0.10; 95% CI 0.01–0.82; P = .03). Studies such as these argue that for the majority of patients with no prohibitive surgical risk, there is little benefit to delaying the operation. However, data for patients with neurologic complications from IE are more nuanced.
Approximately 40% of patients with IE have clinical evidence of a cerebral injury (ischemic, intracerebral hemorrhage, infectious aneurysm, brain abscess, or meningitis). Because they represent a substantial proportion of IE patients, the presence of neurological injury deserves a special evaluation of their mortality risk. In a Japanese cohort, Okita et al. found that among patients who suffered cerebral infarcts, patients who had surgery between 8 and 21 days and more than 22 days after ICH had a lower incidence of in-hospital death (OR 0.79; P = .843 and OR 0.12; P = .200, respectively) compared with those who had surgery within 7 days [ ]. These results contrast with patients who suffered a cerebral infarct; they had higher incidences of hospital death with surgery between 15 and 28 days and after 29 days from CI (OR 5.90; P = .107 and OR 4.92; P = .137). Although statistically insignificant, the authors recommend avoidance of surgery within 7 days of ICH related to IE. Similarly, Piper et al. in a prospective study of 108 patients with IE and thromboembolic stroke found favorable survival for patients operated within 72 h of presentation (76% at 40 months; P < .0001) [ ].
The reason for such a discrepancy in outcomes (whether it is mortality or further neurological injury) based on the timing of surgery is not well understood but can be explained by a selection and observer bias across different studies. Many patients with IE who suffered a stroke, particularly those with existing medical comorbidities or multiple prior cardiac surgeries, have a prohibitive surgical risk and are often not offered valve replacement resulting in a selection bias. Similarly, patients who remain neurologically devastated despite optimal medical therapy are also not offered surgery and often deteriorate clinically. On the contrary, patients who demonstrate neurological recovery or no significant deterioration while undergoing medical management are considered for surgery with better outcomes. Their survival may not be entirely attributable to the length of surgical delay but to a better functional reserve, medical optimization prior to their intervention, or to overall low disease burden. It is unclear how aggressively this last group of patients who are otherwise healthy and low risk was managed surgically in the different cohorts evaluated, possibly introducing an observer bias. Data from the studies reported here do not shed light on the decision process regarding patients’ selection for surgery, an inherent limitation of retrospective studies.
Current societies guidelines including the European Society of Cardiology (ESC) support surgery following a neurologic event with the recommendation that perioperative risk must be balanced against the postoperative prognosis [ ]. ESC recommendations can be summarized as follows:
In the case of silent cerebral emboli or transient ischemic attack (TIA), surgery should proceed without delays if indicated (Class I, level of evidence B) [ ].
Cardiac surgery is not contraindicated following ischemic stroke unless the neurological prognosis is too poor [ ].
Optimal time interval between stroke and cardiac surgery is conflicting but recent data favor early surgery [ , ].
In patients with ICH, surgery should be postponed for at least 1 month (Class IIa, level of evidence B) [ , , ].
If cerebral hemorrhage has been excluded by cranial computed tomography (CT) or magnetic resonance imaging (MRI) and patient is not neurologically devastated (i.e., comatose), surgery is indicated for heart failure, uncontrolled infection, abscess, and persistent emboli, and should not be delayed with low neurological risk (3%–6%) (Class IIa, level of evidence B) [ , ].
The 2016 American Association for Thoracic Surgery (AATS) consensus guidelines on the timing of surgery in IE following a stroke state that an operative delay of 3 weeks or more is reasonable among patients with recent ICH (Class IIa) [ ]. AATS recommendations for ideal timing of surgery for IE with stroke are concordant with ESC and can be summarized as follows:
If a cerebral mycotic aneurysm has been diagnosed, treatment and follow-up of the patients should be in close collaboration with neurologic and neurosurgery expertise (Class I, level of evidence C)
In patients with a recent ICH, a delay of operation for three or more weeks is reasonable (Class IIa, level of evidence B)
Earlier surgery is reasonable for patients with nonhemorrhagic strokes and a strong cardiac indication for urgent surgery (Class IIa, level of evidence B)
Patients with large and multiple strokes and severe neurologic symptoms should be carefully evaluated by a neurologist before being offered surgery (Class I, level of evidence B)
For patients with IE and neurologic symptoms and significant ICH, angiography should be considered to rule out mycotic aneurysm (Class IIa, level of evidence B)
The authors’ practice is consistent with both ESC and AATS guidelines and favors early surgical intervention unless patients present with ICH or large disabling stroke with a high risk of hemorrhagic conversion. By offering surgery in the acute stage, the extent of structural destruction of the valve by IE can be minimized.
Accurately predicting the risks of complications from IE can aid in surgical planning; however, few tools are available that reliably predict the risks of embolization or failure of medical therapy. An embolic risk (ER) calculator was developed based on a multicenter French cohort of more than 1000 patients with IE [ ]. In univariate analyses, age, diabetes, atrial fibrillation, previous embolism, vegetation length, and Staphylococcus aureus were identified as predictors of embolic events and confirmed in a validation model that provides the risk of embolization up to 6 months ( Fig. 20.1 ). The predictive accuracy of the calculator was assessed by evaluating the agreement between predicted and observed embolic events which was excellent. These characteristics included in the model have previously been associated with higher risk of embolization [ ], but risk calculator such as these should be used cautiously owing to a detection bias in a retrospective study that investigated only variables routinely measured at the time of the patients’ admission. As the authors of the calculator model highlight, variables such as coagulopathies or prothrombotic state that can increase the risk of embolization were not measured or evaluated in this model. The authors also did not evaluate the role of native versus prosthetic valve or mitral versus aortic valve in their model. Despite some of these limitations, external validation of the ER calculator has been reported in several studies including a Japanese and Filipino cohorts [ , ].