Stroke Associated with Infective Endocarditis. Frequency and Prognostic Implications

Corresponding Author: Juan José Mengual Chirife Department of Neurology, Complex Hospitalari Moisès Broggi, Spain Email: JuanJose.MengualChirifie@san itatintegral.org Abstract: Stroke can be one of the most severe complications of Infective Endocarditis (IE) and it happens in 14% of patients with IE. Therefore, we aimed to review stroke frequency in patients diagnosed with IE in our hospital from 2010 to 2020. We defined it as IE-Associated Stroke (IEAS) when the stroke happened one month before or after IE's diagnosis. Sixty-six patients were diagnosed with IE during this period time and 15 (22%) suffered IEAS: 13 ischemic strokes, one Subarachnoid Hemorrhage (SAH) and one parenchymal hemorrhage with SAH. Patients with IEAS were more frequently men (80% vs 51% p = 0.042) and had less atrial fibrillation (13.3% vs 37.3%, p = 0.071). Although not statistically significant, Streptococcus sp (53% vs. 44.9%) and Staphylococcus aureus (33.3% vs. 16.3%) were more frequent in patients with IEAS. Mortality was increased in patients with IEAS (60% vs 13.7%, p = 0.001) especially in patients where stroke diagnosis was posterior to IE (88.9% vs 16.7%, p = 0.011). Hypertension (93.8% vs 63%, p = 0.013) and diabetes (53.2% vs 30%, p0 = 0.052) were also more frequent in those patients who died. In multivariate analysis, hypertension (OR 13.7 95% CI 1.4-138.4) and IEAS (OR 12.8 95% CI 2.8-58) were independently associated with death. The frequency of IEAS in our patients is comparable to what has been described in the literature. IEAS was associated with increased mortality.


Introduction
It has gone more than a century since Sir William Osler made the first exhaustive description of Infective Endocarditis (IE) (Levy, 1985) in the XIX century. However, despite the advances in diagnosis and treatment, IE remains a clinical challenge nowadays. Moreover, the complexity of IE makes it necessary to set multidisciplinary "endocarditis teams" as has been recommended by the European Cardiology Society in a consensus statement in 2015 (Habib et al., 2015).
In high-income countries an increasingly older population and the associated comorbidities are major determinants for the development of IE. In addition, degenerative valvular disease, prosthetic valves, intracardiac implantable devices and invasive medical procedures also play a role in developing IE. However, in low-middle income countries, rheumatic heart disease and congenital heart diseases still remain the most important factors (Cahill and Prendergast, 2016;Murdoch et al., 2009;Njuguna et al., 2017).
IE is diagnosed with the modified Duke criteria as definite, possible and rejected (Li et al., 2000). Vascular events, such as major arterial emboli, mycotic aneurysms and intracranial hemorrhages, are included in IE's diagnosis's minor criteria. In this setting, stroke can appear as an extracardiac complication of IE in 14% of the cases (García-Cabrera et al., 2013) and usually carries a poor clinical prognosis (García-Cabrera et al., 2013;Thuny et al., 2007;Morris et al., 2014).  (Cahill and Prendergast, 2016;Hoen and Duval, 2013;Heiro et al., 2000;Bashore et al., 2006;Jones and Siekert, 1989;Greenlee and Mandell, 1973). Accordingly, neurology specialists may be involved and play an essential role in managing patients with IE.
The causal agents of IE are varied, in Table 1 we mentioned the most frequently involved.
Our study aimed to describe stroke frequency during the last ten years in patients diagnosed with IE in our hospital and describe the main clinical characteristics and prognostic implications.

Methods
Moisès Broggi Hospital is located in Barcelona's south metropolitan area and attends a population of 422.000 inhabitants. We selected from our local electronic records patients that had been diagnosed with IE (ICD-10 code I33) and ischemic stroke (ICD-10 codes I63), hemorrhagic stroke (ICD-10 codes I61), or subarachnoid hemorrhage (SAH, ICD-10 codes I60). Finally, we defined IE-associated stroke (IEAS) when stroke happened one month before or after IE diagnosis. We collected the following variables: age, sex, hypertension, diabetes, dyslipidemia, Chronic Kidney Disease (CKD), Ischemic Heart Disease (IHD), Atrial Fibrillation (AF), presence of Implantable Cardiac Devices (ICD), affected valve, microorganism causing IE, valve replacement surgery, the temporal relationship between stroke and IE diagnosis and death. Patients diagnosed with IE were divided into IEAS (IEAS+) patients and those without IEAS (IEAS).
Statistical analysis was performed with IBM SPSS 26. Firstly, we compared patients with and without IEAS. Secondly, we selected those patients with IEAS and compared those in which stroke diagnosis preceded the diagnosis of IE and those with a stroke posterior to IE diagnosis. Continuous variables were described as means and Standard Deviations (SD) and categorical variables were expressed as percentages. Frequency tables were analyzed using Pearson's chi-squared test or Fisher's test, as appropriate. Quantitative differences between groups were compared using Student's t-test, ANOVA, or Mann-Whitney U test, as appropriate. We considered as statistically significant p values of <0.05 and all hypotheses were two-sided. When required, we used multivariate analysis to adjust for confounding variables.

Results
Sixty-six patients were diagnosed with IE in our hospital from 2010 to 2020. The main demographic and clinical characteristics are described in Table 2. Fifteen patients (22.7%) suffered an IEAS: 13 ischemic strokes, one Subarachnoid Hemorrhage (SAH) and one parenchymal bleeding with SAH. In addition, the patient with an isolated SAH underwent a brain angiogram and was diagnosed with a mycotic aneurysm (Fig. 1).
The middle cerebral artery was the arterial territory most frequently involved in patients with ischemic stroke and 10 out of 13 (77%) patients with ischemic stroke had acute ischemic lesions in more than one arterial territory, suggesting the embolic nature.

Discussion
The incidence of IE is 3-9 cases per 100,000 per year (Cao et al., 2018;Hoen and Duval, 2013) and it has been stable over last decades, despite the advance of preventive medicine strategies and better diagnostics tools (Hoen and Duval, 2013;Murdoch et al., 2009). Stroke in the IE setting is infrequent (percentages can vary from 10-50% for embolic events, 65% affecting the central nervous system) (Heiro et al., 2000;Morris et al., 2014) but carries an ominous prognosis. For example, in our series of patients with IE for ten years, stroke appears as a complication of IE in 22.7%. This percentage is similar to previously reported (Thuny et al., 2007;Sotero et al., 2019;Anderson et al., 2003;Fabri Jr et al., 2006;Barsic et al., 2013;Hasbun et al., 2003), but it might be even higher since the risk of stroke remains increased in the first year after the IE diagnosis (Merkler et al., 2015;Østergaard et al., 2019). The observed mortality in our series is also comparable to other studies (García-Cabrera et al., 2013;Anderson et al., 2003;Hasbun et al., 2003;Hobohm et al., 2016).
However, Dickerman et al. (2007) showed us that the stroke rate during the first week of antibiotic therapy was 4.82 of 1000 patient-days. Still, this rate fell by 65% in the second week of treatment to 1.71 of 1000 patient days (p<0.001) and continued constantly falling after the start of antibiotic therapy. This could be explained by the breakdown of the vegetation formed by inflammatory cells, endothelial cells, invading microorganisms, platelets and red blood cells with antimicrobial therapy, leading to a decrease in the vegetation's size and or stabilization. We defined IEAS when the stroke happened either one month before or after the IE diagnosis because the risk is maximal during this period and more likely a causative relationship with IE. Compared to other studies, we found a similar pattern of microorganisms and valvular involvement (Hoen and Duval, 2013;Bashore et al., 2006;Hobohm et al., 2016;Lee et al., 2014), being S. aureus the most common single microorganism responsible for IE. This phenomenon is related to the increased healthcare contact in the general population and the chronification of diseases (Cahill and Prendergast, 2016;Bashore et al., 2006;Jones Jr and Siekert, 1989;Greenlee and Mandell, 1973;Murdoch et al., 2009), but possibly Streptococcus sp. still represents the foremost responsible for IE in the low and middle-income countries with a progressive increment of S. aureus infection (Cahill and Prendergast, 2016;Njuguna et al., 2017). Similarly, we also found the classical multi-territorial involvement described for ischemic stroke caused by IE (Jones Jr and Siekert, 1989;Greenlee and Mandell, 1973;Lee et al., 2014;Singhal et al., 2002), being the middle cerebral artery the most affected territory. In fact, 3 of our patients with ischemic stroke before IE diagnosis had a single territorial infarct. Lee et al. (2014;Singhal et al., 2002) described four patterns of ischemic lesions in DWI MRI (A) single lesion, (B) territorial infarction, (C) disseminated punctate lesions and (D) numerous small (<10 mm) and medium (10 to 30 mm) or large (>30 mm) lesions in multiple territories. It's crucial to highlight that only pattern D was exhibited in Non-Bacterial Thrombotic Endocarditis (NBTE) patients. Therefore, although brain imaging can show different patterns suggestive of an underlying IE, other clinical features such as fever, raised acute phase reactants or the presence of a new cardiac murmur must be taken into account.
The main result of our study is that stroke is associated with significantly increased mortality in patients with IE, especially when IE diagnosis preceded the stroke. Similar findings were reported by Epaulard et al. (2009). In a small study with 32 patients with IE, mortality was higher in patients with stroke after IE diagnosis (Murdoch et al., 2009). Patients with acute stroke had probably been admitted to stroke units, with early neuroimaging and cardiac workout that had allowed an earlier detection and treatment of IE. However, the difference in mortality between patients diagnosed with stroke before or after IE should be confirmed with a larger sample. This association of IEAS with mortality was independent of hypertension, another factor associated with increased mortality and probably the fatal evolution might be caused by stroke occurrence.
This study has significant limitations; apart from the limited number of patients and its retrospective nature, we cannot assure that the patients had been treated with coordinated protocols. This might impact mortality, for instance, the lack of cardiac surgery service inwards. However, we think that the important difference in mortality between both groups makes this point unlikely. Another caveat is that only patients with focal neurological signs were evaluated by a neurologist and went under a selection bias neuroimaging. It's well known that silent ischemic lesions are detected with routine neuroimaging, preferably cerebral MRI, but the ideal time and clinical relevance are still unknown and need further evaluation.
In conclusion, Stroke is an infrequent but serious complication of IE, associated with a high mortality, especially when stroke happens after IE diagnosis. The causing microorganisms are similar to previous series. A high grade of clinical suspicion and early diagnosis of IE might help to improve clinical prognosis in these patients.