Valve surgery in active endocarditis patients complicated by intracranial haemorrhage: the influence of the timing of surgery on neurological outcomes

Abstract

OBJECTIVES

The risk of neurological deterioration during valve surgery using cardiopulmonary bypass under systemic heparinization in infective endocarditis (IE) patients with intracranial haemorrhage (ICH) is unknown. The objective of this retrospective study was to investigate the stratified risk related to the timing of valve surgery on neurological outcomes in patients with active IE and preoperative ICH.

METHODS

From 2004 to 2012, 246 patients underwent valve surgery for IE in hospitals enrolled in the Osaka Cardiovascular Research Group. Of these, a group of 30 patients had preoperative ICH, and they included 18 patients with cerebral haemorrhage, 8 with subarachnoid haemorrhage and 4 with haemorrhagic infarction. The preoperative characteristics, neurological statuses and postoperative results of these patients were retrospectively explored to analyse the effects of the timing of surgery on neurological outcomes.

RESULTS

Twenty-one patients had symptomatic ICH, and the median modified Rankin score was 1.5 (95% confidence interval [CI] 1.2–2.8). Eight patients were diagnosed with mycotic aneurysms, and 7 of these patients underwent aneurysm resection or clipping before valve surgery. All 30 patients underwent valve surgery, and the median interval between ICH onset and surgery was 22.5 (95% CI 15.5–39.4) days. Four patients died of multiple organ dysfunction or heart failure. The interval between ICH onset and valve surgery was within 7 days for 5 cases, between 8 and 14 days for 6, between 15 and 28 days for 9 and >29 days for 10. Postoperative neuroimaging showed that neither neurological deterioration nor exacerbation of haemorrhagic lesions had occurred among the 30 patients, regardless of the timing of surgery. However, 2 cases who underwent valve surgery 8 and 81 days after the onset of ICH developed new ectopic asymptomatic haemorrhages postoperatively.

CONCLUSIONS

The risk of postoperative neurological deterioration resulting from the exacerbation of haemorrhagic lesions seemed relatively low, even in IE patients who underwent valve surgery within 2 weeks of ICH onset. However, further evaluation of the sizes and aetiologies of haemorrhagic lesions is vital to establish a safe interval between the ICH onset and surgery.

INTRODUCTION

Infective endocarditis (IE) often causes cerebral complications, bacterially induced cerebral infarction being the most common, followed by cerebral haemorrhage [1, 2]. Despite advances in medical and surgical methods to treat IE patients, there remains a significant mortality associated with these complications [3]. The guidelines from the American Heart Association and the Society of Thoracic Surgeons recommend delaying surgery in IE patients by at least 4 weeks from the onset of cerebral complications [4, 5]. However, several recent reports have noted that patients with preoperative cerebral infarction have a relatively low risk of postoperative neurological deterioration caused by the haemorrhagic transformation of existing cerebral infarctions [6–8].

Nonetheless, these results cannot be applied to patients who have complications associated with preoperative intracranial haemorrhage (ICH) because of the substantial risks associated with cardiopulmonary bypass (CPB) procedures, including extension of the haemorrhage. Furthermore, IE patients with ICH are often contraindicated for surgery, the number of patients who undergo surgery is extremely limited and the associated mortality rate is particularly high at 50% [2]. Therefore, the impact of surgery on neurological outcomes in these patients is uncertain.

The purpose of this retrospective study was to investigate the impact of valve surgery on neurological outcomes in patients with active IE complicated by preoperative ICH.

PATIENTS AND METHODS

Patient selection

This was a retrospective study, and the protocol was approved by the institutional review board of each participating hospital as well as the ethics committee of Osaka University Hospital. Using the Osaka Cardiovascular Research (OSCAR) Group database, 246 patients who had undergone valve surgery for active IE from 2004 to 2012 were identified. Active endocarditis was defined as definitive endocarditis that required antibiotic therapy until surgery. Each patient satisfied the modified Duke criteria for IE [9]. The exclusion criteria included IE that was cured and right-sided endocarditis that did not involve the left-side valves. Four patients did not undergo preoperative brain computed tomography (CT) scan because of cardiogenic shock. Of the 242 patients who underwent preoperative brain CT, 30 were diagnosed with an ICH based on the CT findings. The preoperative characteristics and the neurological statuses of these 30 patients, together with their postoperative results, were analysed.

Echocardiography and laboratory data

All patients underwent transthoracic echocardiography examinations. Vegetation size was determined as the maximum length. Laboratory data from samples taken at the time of IE diagnosis were available for analysis.

Definition of cerebral lesions

Preoperative ICH included intracerebral haemorrhages, haemorrhagic infarctions and subarachnoid haemorrhages (SAHs). An ICH was defined as an area of high density in the head revealed by CT and was confirmed by a radiologist at the participating hospital and an experienced neurologist. The onset of ICH was determined based on the timing of the occurrence of neurological symptoms, and that of ICH in patients without neurological symptoms was determined based on the timing of the CT scan. The size of the cerebral haemorrhage was determined at its maximum length. In this study, the size of a haemorrhagic infarction was determined as the size of the low-density infarct region. Intracranial small, low-density lesions in gradient-echo images obtained using magnetic resonance imaging (MRI) were considered areas of microbleeding [10] and were not considered ICH if the lesions were not evident within the high-density areas on the CT images. The presence of an intracranial mycotic aneurysm was diagnosed based on the results of CT angiography (CTA) or magnetic resonance angiography (MRA). Subsequent angiography was performed based on the decisions made by the physicians at each hospital. The decision to clip or resect the aneurysms depended on the neurosurgeon. Existing neurological symptoms were determined from the hospital records.

Postoperative neurological deterioration was defined as the development of a new neurological deficit that lasted at least 24 h. Postoperative transient convulsions and delayed recoveries from anaesthesia without focal neurological deficits were not considered postoperative neurological deterioration. Twenty-nine of the 30 patients underwent postoperative neurological imaging using CT or MRI during hospitalization. One patient did not undergo postoperative neurological imaging because the patient died 3 days after surgery. Enlargements of pre-existing haemorrhagic lesions or the occurrence of new ectopic lesions on postoperative CT or MRI images were evaluated. During postoperative neuroimaging examinations, ICH included cerebral bleeding, SAHs and subdural haemorrhages, which were evident as high-density areas on CT scans or low-density areas on T₂-weighted MRI images.

Valve surgery

The timing of surgery was based on the published guidelines [4, 5] and was determined by the attending surgeon in each case. All surgical records were reviewed, and all patients underwent valve surgery under CPB with the administration of systemic heparinization for an activated clotting time of >380 s. The body temperature during CPB was determined depending on each surgeon and was confirmed to vary from 20 to 37°C depending on the procedure. For valve replacement, the decision to use a bioprosthetic valve or a mechanical valve mainly depended on the patient's age, as well as the interval between ICH onset and surgery. Two patients who underwent mechanical valve replacement were administered oral warfarin and antiplatelet therapy after extubation. These 2 patients had enough interval since haemorrhagic event until valve surgery (58 and 90 days, respectively), they had anticoagulation therapy immediately after extubation and had no adverse effect of anticoagulation. The decision of anticoagulant therapy in patients who underwent bioprosthetic valve replacement depended on each physician.

Statistical analysis

Continuous variables are shown as medians and 95% confidence intervals (CIs). Categorical variables are reported as frequencies. All statistical analyses were performed using JMP 10.0 (SAS, Inc.).

RESULTS

The preoperative clinical characteristics of the 30 patients are summarized in Table 1. In 25 (83%) patients, blood cultures were positive for Staphylococcus aureus in 13 (43%, 95% CI 30–63%) patients and Streptococcus spp. in 12 (40%, 95% CI 25–58%) patients. The prevalence of S. aureus in these patients was higher than that in other IE patients without ICH [8]. The median vegetation size was 13.0 (95% CI 10.6–15.1) mm in diameter, and 9 (30%) patients had a vegetation of >15 mm. The laboratory variables of these 30 patients were similar to those of those patients with preoperative acute cerebral infarctions reported in our previous study [8].

The patients' preoperative neurological characteristics are presented in Table 2. Twenty-one (70%) patients had neurological symptoms, and of these, 11 had paralysis of the extremities, 6 experienced disturbances in consciousness and 5 had constructional apraxia. The patients' preoperative neurological status was determined according to a modified Rankin Score [11]. The median modified Rankin score was 1.5 (95% CI 1.2–2.8) in these 30 patients. Eight (27%) patients had a score of 0, 7 (23%) had a score of 1, 1 (3%) had a score of 2, 3 (10%) had a score of 3, 7 (23%) had a score of 4 and 4 (13%) had a score of 5. Eighteen patients had a cerebral haemorrhage, and the median diameter was 20 (95% CI 15.5–39.4) mm. SAHs were found in 8 patients, and haemorrhagic infarctions in 4.

Preoperative MRA or CTA images showed mycotic aneurysms in 8 patients. One of the 8 patients with a mycotic aneurysm had it surgically resected, and 6 patients had their aneurysms clipped before valve surgery.

All the 30 IE patients underwent valve surgery, and the median interval between ICH onset and surgery was 22.5 (95% CI 15.5–39.4) days (Table 3). The interval between ICH onset and valve surgery was within 7 days for 5 patients (Fig. 1A), between 8 and 14 days for 6 (Fig. 1B), between 15 and 28 days for 9 and >29 days for 10 (Fig. 1C). Four patients who underwent surgery 4, 25, 31 and 47 days after ICH onset died after surgery of cardiac failure or multiple organ dysfunction. No deaths occurred from cerebrovascular complications. Of the patients who died, 2 underwent the Bentall procedure and redo mitral valve replacements with bioprosthetic valves, 1 underwent a repeat Bentall procedure that was concomitant with coronary artery bypass graft (CABG) surgery and 1 underwent double valve replacements. A 69-year old woman who underwent aortic valve surgery 58 days after ICH onset and a 24-year old man who underwent mitral valve surgery 90 days after ICH onset had mechanical valve replacement. Fifteen patients underwent bioprosthetic valve surgery (2: aortic; 6: mitral; 2: double valve; 3: Bentall procedure; 2: Bentall procedure and mitral), and 6 of survived 11 patients had oral anticoagulation after confirmation of no enlargement of haemorrhagic lesion.

Regarding postoperative neurological function, none of the patients experienced neurological deterioration or died of neurological complications, irrespective of the timing of surgery. One patient who died 3 days after undergoing the Bentall procedure concomitant with CABG surgery did not undergo postoperative neuroimaging; however, this patient showed no postoperative focal neurological dysfunction.

The remaining 29 IE patients underwent postoperative neuroimaging, regardless of their symptoms. Postoperative neuroimaging did not reveal any enlargements of the haemorrhagic lesions, although new ectopic haemorrhages developed in 2 patients. One of these patients, a 52-year old man with a parenchymal cerebral haemorrhage in the left occipital lobe that was 15 mm in diameter, who had an aortic valve replacement with a bioprosthetic valve 8 days after ICH onset developed an asymptomatic focal SAH, although he was not administered postoperative anticoagulation therapy. The other patient, a 49-year old woman, who had a parenchymal cerebral haemorrhage in the left frontal lobe that was 50 mm in diameter and had a mitral valve replaced by a bioprosthetic valve 81 days after ICH onset developed a small parenchymal haemorrhage in the occipital lobe and a small subdural haematoma in the right temporal lobe.

Of the 8 IE patients who had complications associated with preoperative mycotic aneurysms, 7 underwent surgical resection or clipping before valve surgery. These 8 patients underwent valve surgery at a median interval of 20.5 (95% CI 18.0–42.0) days after ICH onset, and none of them developed neurological deterioration after valve surgery. One patient, who did not undergo aneurysm clipping, had a mitral valve replaced 22 days after the onset of a focal SAH. This patient's mycotic aneurysm spontaneously disappeared after antibiotics were administered pre- and postoperatively.

The details of the 11 IE patients who underwent valve surgery within 2 weeks of ICH onset are summarized in Table 4. Most of these patients required early surgery, because they had a high embolic risk, refractory infection or refractory heart failure. Of these patients who underwent early surgery, 2 had preoperative mycotic aneurysms detected by CTA or selective angiography and underwent aneurysm clipping followed by valve surgery, 10 did not deteriorate neurologically and 1 developed a postoperative ectopic focal SAH.

DISCUSSION

Neurological complications occur in 10–40% of patients with IE and are associated with increased mortality [1, 4, 12, 13]. Neurological manifestations related to IE include bacterially induced cerebral infarctions, cerebral infarctions with haemorrhages transient ischaemic attacks, ruptured mycotic aneurysms, meningitis, seizures, headaches, non-focal encephalopathy and peripheral neuropathies.Of these, cerebral infarction without haemorrhage is the most common, followed by cerebral haemorrhage [4–6].

The appropriate timing of surgery for IE patients with cerebral complications remains controversial. Guidelines from the Society of Thoracic Surgeons recommend delaying surgery by at least 4 weeks when patients have experienced major ischaemic strokes or have ICH [4, 5]. On the other hand, they also state that early surgery is reasonable if there are indications for this, including a decline caused by heart failure or recurrent embolisms in patients with small cerebral infarctions without haemorrhages. Recently, several studies have reported that the risk of neurological impairment is lower than estimated in IE patients with cerebral infarctions who undergo early surgery, except in those without haemorrhagic lesions [6–8]. In these studies, systemic heparinization during surgery was rarely found to cause brain oedema or haemorrhagic transformations related to infarctions. However, in these studies, the risks of neurological deterioration or recurrent bleeding in haemorrhagic lesions in patients who underwent early surgery were not investigated.

The guidelines recommend that surgery be delayed for at least 4 weeks, especially for patients with IE complicated by ICH [4, 5]. However, these recommendations are not well founded because of the paucity of reported cases. Even the largest series [3] only included 34 IE patients with ICH, and the majority of valve replacements (n = 21) were performed >4 weeks after the onset of intracerebral haemorrhages. Patients with IE complicated by ICH sometimes have clear indications for early surgery, such as severe infection or progressive heart failure, and when it seems almost impossible to treat them with medical therapy alone; thus, they often have to receive palliative care or undergo surgery, which carries the risk of neurological deterioration. Consequently, it is necessary to clarify the risk of neurological impairment caused by early surgery and then to stratify the influence of the timing of surgery on neurological outcomes to achieve better prognoses for IE patients with ICH-associated complications.

We reviewed the literature to evaluate the risk of early surgery on neurological outcomes and found 89 IE patients with preoperative ICH [1, 3, 7, 12, 14–23] (Table 5). Of these, cases in which the timing of surgery was not documented were excluded from our analysis. We found that 9 patients had undergone surgery within 2 weeks of ICH onset and, of these, 1 developed neurological deterioration following surgery [3]. A further 15 patients underwent surgery between 15 and 28 days after ICH onset and none of them showed neurological deterioration, and 29 underwent surgery >29 days after ICH onset and neurological deterioration was observed in only 1 patient. While the risk of postoperative neurological deterioration in patients with preoperative ICH may not be as high as previously thought, there are shortfalls in this thinking. For example, patients who could not undergo valve surgery may not have been included in the analyses, and those who developed neurological deterioration postoperatively might not have been reported in these papers.

In the present study, none of the 11 patients who underwent surgery within 2 weeks of ICH onset developed postoperative neurological deterioration, although 1 (9%) patient developed a small postoperative asymptomatic subdural haematoma. Five patients underwent surgery within 7 days of ICH onset, and they showed neither neurological deterioration nor postoperative ICH. Nevertheless, we cannot conclude that surgery within 7 days is safe because the number of patients who underwent surgery within 7 days was small and these 5 patients had relatively small haemorrhagic lesions (defined as small haemorrhages within infarctions, focal SAHs or intracerebral haemorrhages 10 mm in diameter). However, the risk of recurrent bleeding seems to be relatively low even in the acute phase, for example, at 1–2 weeks following ICH diagnosis, if the patient has a relatively small haemorrhagic lesion. Our literature review suggests that the risk of neurological deterioration is also relatively low in patients who proceeded to undergo valve surgery in the acute phase. These results agree with our findings that the prevalence of postoperative neurological deterioration is lower than previously estimated.

The relationship between ICH and the administration of anticoagulation therapy is not well established. Diverse ICH aetiologies exist, especially in IE patients, and it is important to clarify the detailed aetiology of ICH in IE patients to uncover the reasons underlying the low prevalence of the postoperative exacerbation of haemorrhagic lesions. Since the aetiology of ICH includes the rupture of mycotic aneurysms, pyogenic arteritis and haemorrhagic infarctions, the risk of postoperative neurological deterioration, and consequently the safe interval between ICH onset and valve surgery, is likely to be specific to each aetiology.

A mycotic aneurysm is formed from a fragile artery infected by septic emboli that has become plugged [24]. Among the 30 patients in the current study, 8 (27%) had mycotic aneurysms detected preoperatively. Of these, 7 patients had their aneurysms surgically treated before valve surgery. The other patient's mycotic aneurysm spontaneously disappeared after antibiotics were administered pre- and postoperatively. If a patient with ICH is diagnosed with a mycotic aneurysm by CTA or MRA, surgical resection or endovascular coiling before valve surgery may reduce the risk of recurrent postoperative bleeding. However, because only 2 of 7 patients with prior mycotic aneurysm clipping or resection underwent valve surgery within 2 weeks of the onset of haemorrhage in this study, it is difficult to estimate the safety interval between aneurysmal treatment and valve surgery.

In addition, it is important to be reminded, that the necessity of prior surgical treatment of unruptured myoctic aneurysm before valve surgery in IE patients is still controversial. It is also important that mycotic aneurysms are often located in peripheral arteries, and may be too small to be detected by CTA or MRA [25]. If all of our patients had undergone selective angiography, the prevalence of mycotic aneurysms might have been much higher, although it remains unclear whether these small peripheral aneurysms should be treated prior to valve surgery.

Rupture of pyogenic arteritis seems to be a mechanism that causes intracerebral parenchymal haemorrhage in patients without obvious mycotic aneurysms. Masuda et al. [25] reported that parenchymal cerebral haemorrhage caused by pyogenic arteritis was prevalent in patients with IE at autopsy. It is difficult to detect pyogenic arteritis and to distinguish from undetected mycotic aneurysm precisely at the time of ICH onset in IE patients, because pyogenic arteritis cannot be diagnosed by neuroimaging. It is also difficult to treat ruptured pyogenic arteritis surgically, because other intracerebral arteries are likely to be affected by the condition. In patients with postoperative ectopic bleeding, the presence of pyogenic arteritis should be the aetiology given most consideration. To prevent the recurrence of bleeding or ectopic haemorrhage during valve surgery, we believe that it is important to treat these fragile arteries with sufficient amounts of antibiotics before valve surgery.

Haemorrhagic infarction, a haemorrhagic transformation of an infarcted lesion, is also an important mechanism underlying ICH in IE patients. We have previously reported that such a transformation is relatively rare, even in patients who have experienced major ischaemic strokes and who have undergone surgery within 2 weeks of acute cerebral infarction detection [8]. If haemorrhagic infarction develops preoperatively, then early surgery is thought to carry a low risk of postoperative exacerbation of haemorrhagic lesions if they do not enlarge over the course of several days following surgery.

The present study has several limitations. First, it was a retrospective multicentre study with a small number of patients and the timing of surgery depended on the attending physician. Secondly, patients with major ICH who could not undergo valve surgery because of poor health status were not considered for inclusion in the study. Thirdly, we could not differentiate among the ICH aetiologies precisely, although particular aetiologies may affect the risk of postoperative neurological exacerbation. Finally, while it is reasonable to consider that the risk of postoperative neurological deterioration depends on the size and the aetiology of the preoperative haemorrhage, haemorrhage sizes were relatively small with various aetiologies also especially in patients who proceeded to early surgery. A definitive relationship between size and safety interval could not be detected in this study, although it is an important problem to be solved.

CONCLUSIONS

The risk of postoperative neurological deterioration resulting from the exacerbation of haemorrhagic lesions seemed relatively low, even in IE patients who underwent valve surgery within 2 weeks of ICH onset, if the haemorrhagic lesion was small. However, further evaluation of the sizes and aetiologies of haemorrhagic lesions is vital to establishing a safe interval between the ICH onset and surgery.

ACKNOWLEDGEMENTS

The authors thank the doctors of the OSCAR group for their help in identifying the patients for the study and the preparation of their medical records.

Conflict of interest: none declared.

REFERENCES