Clinical and Imaging Characteristics of Infective Endocarditis‐Associated Versus Atrial Fibrillation‐Associated Stroke

Authors: Taewoo Kim, Kwanju Song, Hyunsun Oh, Jeong Yoon Song, Kyubong Lee, Byung Joo Sun, Wookjin Yang, Jun Young Chang, Dong‐Wha Kang, Sun U. Kwon, Bum Joon Kim

Abstract

Background

Infective endocarditis (IE) is a major cause of cardioembolic stroke, leading to poor outcomes if not diagnosed appropriately. We compared clinical and imaging characteristics between IE‐associated stroke (IE‐stroke) and atrial fibrillation stroke presenting with fever.

Methods

Patients with embolic infarction diagnosed with IE based on transthoracic echocardiography were reviewed. Patients with atrial fibrillation‐stroke presenting with fever served as the comparator group. Clinical characteristics, imaging characteristics, and outcomes were compared. Lesion patterns were analyzed according to the number of lesions and vascular territories. Parenchymal hematoma, stroke recurrence, and mortality were also investigated.

Results

The study included 170 patients with IE‐stroke and 153 patients with atrial fibrillation stroke. Those with IE‐stroke demonstrated smaller lesions (21.6±19.3 mm versus 66.9±38.0 mm, P<0.001), a higher proportion with >10 lesions (44.7% versus 5.2%, P<0.001), and predominant bilateral anterior–posterior circulation involvement (54.1% versus 2.6%, P<0.001). Independent predictors of IE‐stroke included younger age (odds ratio [OR], 0.87 [95% CI, 0.82–0.92], P<0.001), smaller lesion size (OR, 0.94 95% CI, 0.92–0.97, P<0.001), and 3‐territory involvement (OR, 81.21 [95% CI, 16.20–407.00], P<0.001). Parenchymal hematoma rates were comparable (12.9% versus 11.1%, P=0.614). Ischemic lesion recurrence was higher in IE‐stroke (53.2% versus 18.1%; OR, 6.94 [95% CI, 3.41–14.12], P<0.001). Although unadjusted mortality rates did not differ, adjusted analysis showed higher odds of 3‐month mortality in IE‐stroke (OR, 3.82 [95% CI, 1.71–8.50], P=0.001).

Conclusions

IE‐stroke lesions were more numerous, smaller, and dispersed across multiple vascular territories with higher recurrence rates compared with atrial fibrillation stroke. Despite lesion size difference, parenchymal hematoma rates were similar. These characteristics may aid in distinguishing IE‐stroke in patients with embolic stroke and fever.

Clinical PerspectiveWhat Is New? Infective endocarditis‐associated stroke demonstrates a distinctive neuroimaging phenotype characterized by smaller, multifocal lesions with bilateral anterior–posterior circulation involvement, coupled with significantly higher rates of systemic embolism and 1‐month stroke recurrence compared with atrial fibrillation stroke.Despite the presence of smaller lesions and lower rates of large vessel occlusions, infective endocarditis‐associated stroke exhibits comparable parenchymal hematoma frequency indicating a hemorrhagic tendency likely attributable to the septic nature of emboli. What Are the Clinical Implications? Brain imaging characteristics in patients presenting with embolic stroke and fever may serve as early diagnostic markers for differentiating infective endocarditis from atrial fibrillation, guiding targeted cardiac evaluation strategies and therapeutic decision making in this high‐risk population.

Fever is a common occurrence among patients with stroke, with ∼25% experiencing fever at the onset and up to 50% developing it during hospitalization. ^1^ , ^2^ Various factors can contribute to fever, including pneumonia, urinary tract infection, and infective endocarditis (IE), which is a direct cause of ischemic stroke. IE is a well‐recognized infectious cause of ischemic stroke and is associated with embolic stroke in ∼20% of cases due to septic emboli. ^3^ Additionally, atrial fibrillation (AF) is the most prevalent cause of embolic stroke and may initially present with fever due to concurrent infection. ^4^ Therefore, both IE‐ and AF‐related strokes should be considered primary diagnoses in patients presenting with embolic stroke and fever.

The diagnosis of IE primarily relies on the modified Duke criteria. ^5^ Key components of the diagnosis include identifying the causative microorganism through blood cultures and confirming the presence of vegetations via echocardiography, particularly transesophageal echocardiography (TEE). However, blood cultures require time to yield results, and the use of TEE is limited in cases of acute ischemic stroke due to the potential increased risk of hemorrhagic transformation (HT). Despite these challenges, it is essential to quickly differentiate between IE and other sources of embolism, as their treatments differ significantly. Antibiotic therapy is the cornerstone of IE treatment, whereas the use of anticoagulation remains controversial due to the risk of hemorrhagic complications. ^6^ In contrast, anticoagulation is the preferred treatment for AF‐related stroke, with recent studies indicating that early anticoagulation leads to favorable outcomes. ^7^

Early identifiable characteristics of patients with IE‐stroke include fever and laboratory tests indicating elevated infection markers. ^8^ The lesion patterns observed in ischemic strokes, which represent the underlying mechanisms of the stroke, can also provide valuable insights. ^9^ In our study, we compared patients with IE‐stroke and those with AF‐associated stroke presenting with fever in terms of their clinical, laboratory, and imaging characteristics, including brain lesion patterns, and analyzed the effectiveness of these factors in the early differentiation of IE‐stroke from other embolic strokes accompanied by fever. Additionally, we evaluated clinical outcomes, including HT, ischemic stroke recurrence, and mortality.

METHODS

Patient Selection

This retrospective study screened patients with acute ischemic stroke who were admitted to Asan Medical Center (Seoul, Korea) between January 2012 and May 2024. The inclusion criteria were (1) a diagnosis of acute ischemic stroke within 7 days of stroke onset, (2) age greater than 18 years, and (3) a diagnosis of IE according to the modified Duke criteria. Only patients with definite IE were included in the IE‐stroke group, and those with clinically probable nonbacterial thrombotic endocarditis were excluded. Among the patients who underwent heart valve surgery to treat IE, only those diagnosed with acute ischemic stroke before the surgery were included and those diagnosed with acute ischemic stroke after the surgery were excluded.

As a control group, we included patients who were admitted during the same period and diagnosed with AF and cerebral infarction and had blood cultures obtained within 24 hours of admission due to fever. Fever was defined as a tympanic temperature of 37.5 °C or higher. ^10^ We excluded patients who did not undergo a brain magnetic resonance imaging during hospitalization or those without acute lesions on brain diffusion‐weighted image (DWI). Cases where the cause was determined to be large artery atherosclerosis, small vessel disease, or cancer‐related stroke, rather than AF, as assessed by the stroke neurologist, were excluded. This study was approved by the institutional review board of our center. Informed consent was not obtained due to the retrospective nature of the study. The data that support the findings of this study are available from the corresponding author upon reasonable request.

Clinical and Echocardiographic Data

Patient demographics, clinical characteristics, medical history, and concurrent medications were extracted from the medical records. Baseline laboratory tests conducted on the day of admission included complete blood cell counts, blood urea nitrogen, creatinine, C‐reactive protein, total protein, total bilirubin, prothrombin time‐international normalized ratio, and activated partial thromboplastin time. Blood cultures were obtained from all patients upon admission, and the results were documented.

According to our center’s protocol, patients suspected of having an embolic stroke underwent transthoracic echocardiography within 72 hours of admission. If IE was clinically suspected, such as in cases of embolic stroke accompanied by fever, TEE was also performed, unless the patient was unstable. The presence of structural heart disease was assessed using both transthoracic echocardiography and TEE. Additionally, left atrial size and left ventricular ejection fraction were measured via transthoracic echocardiography. In patients with IE, the infected valve (either mitral or aortic) and the size of the vegetation, determined by its maximal length, were assessed through TEE. If TEE was not feasible, the infected valve and vegetation size were evaluated using transthoracic echocardiography.

Additional imaging workups, such as abdominal computed tomography scans, were frequently performed to evaluate the infection foci, both in patient with suspected IE and AF‐stroke with fever. For patients with IE, the history of valve surgery and postsurgical strokes was also evaluated. Data on acute reperfusion treatments, which could influence clinical outcomes, were also reviewed.

Neuroimaging Study

DWI, time‐of‐flight magnetic resonance angiography (MRA) for the evaluation of intracranial arteries, and contrast‐enhanced MRA for extracranial arteries were performed on all patients using a 3.0T magnetic resonance imaging scanner at the emergency medical center. Additionally, gradient echo or susceptibility‐weighted imaging, T2‐weighted imaging, fluid‐attenuated inversion recovery, and contrast‐enhanced T1‐weighted images were obtained.

DWI lesions were analyzed in terms of their number, size, and distribution across the affected vascular territories, referencing previous studies. ^11^ , ^12^ The size of each lesion was determined by measuring the maximal length of the largest lesion on the axial DWI image. Vascular territories were categorized into left anterior circulation (AC), right AC, and posterior circulation, and the number of affected territories was classified based on lesion distribution. Gradient echo or susceptibility‐weighted imaging images were used to assess whether infarcted lesions had experienced HT, which was further classified as hemorrhagic infarction or parenchymal hematoma (PH).

In patients with IE‐stroke, we also assessed the presence of mycotic aneurysms, meningitis, brain abscesses, and vasculitis. MRA was used to evaluate large vessel occlusions associated with cerebral infarct lesions. The extracranial arteries examination included the proximal internal carotid and vertebral arteries (V1, V2, and V3), and the intracranial arteries assessed comprised the distal internal carotid, middle cerebral (M1 and M2), anterior cerebral (A1 and A2), posterior cerebral (P1 and P2), basilar, and vertebral arteries (V4).

According to our center’s protocol, follow‐up DWI, gradient echo, or susceptibility‐weighted imaging, and MRA were routinely performed on the fifth day of admission, regardless of symptom status, in both groups. If any changes in neurological status occurred, additional DWI was performed to check for new infarctions. Two investigators (T.K. and K.S.), who were blinded to the patients’ clinical information, independently evaluated the imaging results. In cases of discrepancies, a third investigator (B.J.K.) was consulted to resolve the issue.

Outcomes

For patients with large vessel occlusion identified on the initial MRA, recanalization—whether spontaneous or following endovascular therapy—was evaluated using follow‐up MRA performed within a week from admission. Clinical outcomes, including rupture of a mycotic aneurysm, symptomatic stroke recurrence, and mortality, were assessed at 3 months after stroke. Outcomes were obtained either through outpatient clinic visits with a neurologist or via structured telephone interviews conducted by a trained nurse. Symptomatic recurrent ischemic stroke was defined as the appearance of new neurological symptoms or significant worsening of preexisting deficits, accompanied by new ischemic lesions not visible on the initial DWI.

Imaging outcomes included HT and any new ischemic lesion identified on follow‐up DWI. Imaging studies performed within 1 month of stroke onset were used to define these outcomes. Any new ischemic lesion was defined as a new lesion on follow‐up DWI that was absent on the baseline scan, regardless of symptom status. Given the high embolic risk associated with valve surgery, follow‐up DWI performed after valve surgery was excluded from the analysis, as surgery itself may be the direct cause of the new lesion. Hemorrhagic infarction was defined as petechial bleeding within the infarcted area without mass effect, and PH was defined as a dense, space‐occupying hemorrhage within the brain parenchyma, typically associated with mass effect. ^13^ , ^14^

Statistical Analysis

Clinical and imaging characteristics were compared between patients with IE‐stroke and those with AF‐stroke. Continuous variables are presented as means±SDs, and categorical variables are expressed as numbers and percentages (%). The chi‐square test and Fisher’s exact test were used to compare categorical variables, and the independent t test was employed for parametric analysis. Missing data were excluded from analyses.

A multivariable analysis to predict IE was conducted using binary logistic regression. For univariable analysis, we included clinical, laboratory, and neuroimaging characteristics. As the analytical objective was to predict IE at initial presentation, variables related to treatments (intravenous thrombolysis, endovascular therapy) were excluded due to temporal sequence considerations. Among imaging characteristics, redundant variables were excluded to avoid collinearity (specific vascular territory patterns when number of vascular territories was included; lesion length≥15 mm when continuous lesion length was included). All variables from clinical data and imaging characteristics with a P value of <0.05 in univariable analysis were included in the multivariable analysis. Odds ratios (ORs) and 95% CIs were calculated. Multicollinearity was also assessed when selecting variables for the multivariable analysis. Multicollinearity was identified if the variance inflation factor exceeded 10 or if Pearson’s correlation coefficient for continuous variables exceeded 0.9. For categorical variables, a Cramer’s V >0.4 indicated a relatively strong association. ^15^

For outcome analysis, we performed both crude and adjusted analyses using logistic regression. The primary adjusted models included age, sex, hypertension, dyslipidemia, coronary artery disease, and body temperature. Extended multivariable models encompassed all primary covariates and additionally included white blood cell count, hemoglobin, platelet count, creatinine, total protein, C‐reactive protein, prothrombin time‐international normalized ratio, activated partial thromboplastin time, largest lesion length, and a number of vascular territories as covariates. These variables were selected based on their significant differences between groups in baseline characteristics and their potential clinical relevance to outcomes. All statistical analyses were conducted using SPSS software version 27.0 (IBM Corp., Chicago, IL), and a P value of <0.05 was considered statistically significant.

RESULTS

Demographics and Clinical Data

During the study period, 11 319 patients with acute ischemic stroke were admitted to our center, of whom 2076 (18.3%) were classified as having cardioembolic strokes. Among these, 264 (12.7%) patients were initially identified as having IE‐stroke. Of them, 39 (15.4%) patients did not meet the modified Duke criteria for definite IE, 26 (10.2%) experienced strokes following valve surgery, 17 (6.7%) lacked appropriate brain imaging, and 12 (4.7%) were ultimately diagnosed with nonbacterial thrombotic endocarditis. In total, 170 patients with IE‐stroke were finally included.

As a comparator group, we screened 330 patients with ischemic stroke with AF and blood cultures obtained within 24 hours of admission. Among them, 81 (24.5%) lacked appropriate imaging findings, 51 (15.5%) had other identified causes of stroke, and 45 (13.6%) presented with fevers below the study’s criteria. Ultimately, 153 patients were included in the AF‐stroke comparator group (Figure 1).

IE‐specific characteristics are summarized in Table 1. Staphylococcus aureus was the most prevalent microorganism, identified in 51 (30%) patients, followed by viridans streptococci and coagulase‐negative staphylococci. In 152 (89.4%) patients, the organism was detected through initial blood cultures; in 6 (3.5%) patients, it was identified via postoperative biopsy, whereas no microorganism was found in 12 (7.1%) patients. Mitral valve infection was present in 91 (53.5%) patients, aortic valve infection in 62 (36.5%) patients, and both aortic and mitral valve infections in 17 (10.0%) patients. Prosthetic valve endocarditis was diagnosed in 53 (31.2%) patients, whereas the remaining 117 had native valve endocarditis. In addition to embolic stroke, mycotic aneurysms, a common central nervous system complication, were identified in 17 (10.0%) patients. Of the 170 patients, 109 (64.1%) underwent valve surgery during their hospital stay for IE, and recurrent ischemic stroke occurred in 18 (16.5%) patients post surgery.

Clinical and Laboratory Characteristics

Table 2 shows the comparison of clinical and laboratory factors between the groups with IE‐stroke and AF‐stroke. The group with IE‐stroke had a younger age (62.1±16.4 versus 74.3±9.6 years, P<0.001) and had a higher proportion of male patients (57.6% versus 45.1%, P=0.024). Conventional risk factors, such as hypertension (47.6% versus 76.5%, P<0.001), dyslipidemia (27.1% versus 39.9%; P=0.015), and a history of coronary artery disease (9.4% versus 17.6%, P=0.030) were more prevalent in the group with AF‐stroke. Prior heart valve surgery (35.3% versus 6.5%, P<0.001) and the presence of prosthetic valves were more common in the group with IE‐stroke. Intravenous thrombolysis (0% versus 9.2%, P<0.001) and endovascular treatment (2.9% versus 15%, P<0.001) were performed more frequently in the group with AF‐stroke. Systemic embolism was more prevalent in the group with IE‐stroke (50.9% versus 12.8%, P<0.001).

In terms of laboratory results, the group with IE‐stroke showed higher values of white blood cell count (13.0±6.4 versus 10.8±4.1 10^9^/L, P<0.001) and C‐reactive protein (12.3±3.4 versus 9.2±6.4 mg/L, P<0.001) and lower values of hemoglobin levels (11.0±2.2 versus 13.1±2.2 g/dL, P<0.001) and platelet count (159.9±94.5 versus 213.1±68.5, 10^9^/L P<0.001). Additionally, the group with IE‐stroke had a higher left ventricular ejection fraction (59.5±9.4 versus 56.9±10.6, P=0.022) and a smaller left atrial diameter (42.9±7.9 versus 47.3±8.3 mm, P<0.001). Blood cultures were conducted in both groups, revealing positive results in 152 (89.4%) patients in the group with IE‐stroke and 3 (2.0%) patients in the group with AF‐stroke.

Imaging Characteristics

Table 3 summarizes the brain imaging characteristics of the groups with IE‐stroke and AF‐stroke. The patterns of brain lesions differed significantly between the 2 groups. IE‐stroke lesions were more frequently found in bilateral ACs (15.3% versus 6.5%, P=0.013) and in both bilateral ACs and posterior circulation (54.1% versus 2.6%, P<0.001), whereas AF‐stroke lesions were more frequently associated with unilateral AC (18.2% versus 62.1%, P<0.001) and posterior circulation (4.1% versus 22.2%, P<0.001) locations. The number of vascular territories involved also varied significantly between the groups (χ^2^=138.01, P<0.001). IE‐stroke was strongly associated with multiple lesions across three vascular territories (54.1% versus 2.6%), whereas AF‐stroke was associated with single‐vascular‐territory lesions (22.4% versus 85.0%).

Although the group with IE‐stroke had a higher number of lesions compared with the group with AF‐stroke, the sizes of the lesions were significantly smaller in the group with IE‐stroke (21.6±19.3 versus 66.9±38.0 mm, P<0.001). Additionally, larger vessel occlusions were less common in the group with IE‐stroke (10.6% versus 65.1%, P<0.001). The initial HT rate was similarly high in both groups, ∼20%, with no significant difference observed between them.

Multivariable Analysis

As a model for early differentiation between IE‐stroke and AF‐stroke in patients with embolic stroke presenting with fever, we conducted a multivariable analysis (Table 4). In this analysis, the following factors were found to be independently associated with IE‐stroke: age (OR, 0.87 [95% CI, 0.82–0.92], P<0.001), hemoglobin (OR, 0.49 [95% CI, 0.36–0.65], P<0.001), C‐reactive protein (OR, 1.10 [95% CI, 1.02–1.18], P=0.019) lesion size (OR, 0.94 [95% CI, 0.92–0.97], P<0.001), and the number of vascular territories (OR, 81.21 [95% CI, 16.20–407.00], P<0.001 for 3 vascular territories versus 1 vascular territory).

Outcomes

Recanalization rates were comparable between the 2 groups (44.4% versus 59.6%, P=0.232), with no significant difference persisting after adjustment (OR, 0.40 [95% CI, 0.11–1.48], P=0.169). Symptomatic recurrence of ischemic stroke was noted higher in the group with IE‐stroke (n=26, 15.3%) than in patients with AF‐stroke (n=7, 4.6%; P=0.002); adjusted logistic regression analysis confirmed this association (OR, 3.71 [95% CI, 1.22–11.34], P=0.021). Three‐month mortality rates were high in both groups, reaching 16.9% in the group with IE‐stroke and 10.7% in the group with AF‐stroke (P=0.147). Adjusted analysis revealed significantly higher mortality risk in patients IE‐stroke (OR, 3.82 [95% CI, 1.71–8.50], P<0.001). HT within 1 month occurred more frequently in the group with AF‐stroke (41.8% versus 59.2%, P=0.002). Adjusted analysis confirmed this inverse association (OR, 0.46 [95% CI, 0.28–0.74], P=0.001). However, extended multivariable analysis incorporating laboratory and imaging parameters, including largest lesion length, showed no significance of this association (OR, 1.44 [95% CI, 0.62–3.35], P=0.402, Table S1). The incidence of PH showed no significant difference between groups (12.9% versus 11.1%, P=0.614). The occurrence of any new ischemic lesions within 1 month was significantly higher in the group with IE‐stroke (53.2% versus 18.1%, P<0.001). Logistic regression analysis also showed higher recurrent embolic risk in the group with IE‐stroke (OR, 6.94 [95% CI, 3.41–14.12], P<0.001), compared with AF‐stroke. (Table 5).

DISCUSSION

The aim of this study was to compare the characteristics of IE‐stroke and AF‐stroke, as well as to investigate their outcomes. Our results indicated that, compared with AF‐stroke with fever, IE‐stroke was associated with a greater number of multiple lesions distributed across 2 or more vascular territories, smaller lesion sizes, and less occlusion of the corresponding artery. The overall rate of any HT within 1 month was higher in AF‐stroke and primary adjusted multivariable analysis maintained this association. However, extended multivariable analysis incorporating comprehensive laboratory and imaging parameters, including largest lesion length, showed no significance of this association (Table S1). Conversely, symptomatic recurrent stroke and any asymptomatic ischemic lesion at follow‐up was significantly more frequent in IE‐stroke. In adjusted analyses, the risk of these recurrent ischemic strokes remained markedly higher in IE‐stroke than AF‐stroke. In multivariable analysis, a greater number of affected vascular territories and smaller lesion sizes were independently associated with IE‐stroke.

The differences in brain lesion patterns between IE‐stroke and AF‐stroke likely stem from the distinct composition of their embolic sources. ^16^ In AF‐stroke, thrombi from the left atrium consist primarily of red blood cells, fibrin, platelets, and other substances. ^17^ , ^18^ Thrombi in AF‐stroke are generally more amenable to recanalization with endovascular therapy and recombinant tissue plasminogen activator due to their higher red blood cell content. ^19^ In contrast, IE‐stroke emboli are a mixture of fibrin, platelets, leukocytes, and red blood cell debris, along with dense clusters of bacteria. ^20^ , ^21^ The presence of bacterial clusters in IE‐stroke emboli may diminish their cohesiveness, resulting in the dispersion of small emboli and the formation of small scattered ischemic lesions across multiple vascular territories, without a large vessel occlusion. Similarly, systemic embolism, which is common in IE, frequently affects the liver, spleen, and lower extremities, further aligning with the multiple scattered lesion pattern characteristic of IE‐stroke. ^16^

IE‐stroke is known to have a high risk of recurrent embolism. Although the risk decreases after the initiation of antibiotic therapy, it remains ∼20%. ^22^ , ^23^ , ^24^ In our study, we observed a significantly higher symptomatic recurrence rate and any new ischemic lesion at follow‐up in the group with IE‐stroke. The low cohesiveness of the vegetation likely contributed to the ease with which septic emboli could dislodge. In addition, the high incidence of any new ischemic lesion at follow‐up may be attributed to the inclusion of strokes identified before the commencement of antibiotic therapy and the use of a 3.0T magnetic resonance imaging scanner in comparison to computed tomography scans. Therefore, the early initiation of antibiotic therapy, and when indicated, early surgical intervention, is crucial for preventing recurrent embolic events. ^25^ , ^26^

In IE‐stroke, intracranial hemorrhage is a common neurological complication, with HT following ischemic stroke and ruptured mycotic aneurysms being the primary causes. ^27^ AF‐associated strokes also frequently result in HT, with a reported incidence of 11.0% among patients with AF‐stroke. ^14^ In comparison, the overall rate of HT in our study was 59.2%, likely due to the larger ischemic lesion sizes observed in the group with AF‐stroke. Because lesion size is the strongest predictor of HT and PH, ^14^ it is understandable that HT were significantly higher in our group with AF‐stroke. Although total HT within 1 month appeared more frequently in AF‐stroke, extended multivariable models showed no significant group difference in total HT, suggesting that the apparent difference was attributable to lesion size. This difference in lesion size may be attributed to the fact that we selected patients with fever, which is often associated with aspiration pneumonia—a condition more prevalent in patients with larger lesions and more severe symptoms. ^2^

The average lesion size in the group with AF‐stroke was larger than that in the group with IE‐stroke. Despite having smaller lesions, the group with IE‐stroke exhibited high rates of HT (41.8%) and PH (12.9%). Notably, PH occurred in 31.0% of HTs in the group with IE‐stroke, compared with 18.9% in the group with AF‐stroke, with PHs in the group with IE‐stroke being more clinically significant due to their association with mortality and disability. ^28^ The higher rates of HTs and PHs observed in our patients with IE‐stroke align with the known hemorrhagic tendency in IE. A previous study also demonstrated that cerebral microbleeds were significantly more frequent in patients with IE compared with controls. ^29^ The higher bleeding tendency in the group with IE‐stroke likely arises from a different pathophysiological mechanism compared with AF‐associated strokes. Although reperfusion injury is a major contributor to HT in AF‐stroke, as evidenced by a large vessel occlusion rate of 65.1% with a subsequent recanalization rate of 59.1%, the group with IE‐stroke had only a 10.6% rate of large vessel occlusion and minimal recanalization. ^30^ , ^31^ Histopathological studies in IE have shown that pyogenic arteritis caused by septic emboli, rather than typical ischemia–reperfusion injury, is the primary cause of hemorrhage in IE‐stroke, which accounts for the high bleeding tendency despite the presence of smaller lesions. ^32^

In addition to differences in imaging, the groups with IE‐stroke and AF‐stroke exhibited distinct baseline characteristics. Patients with IE‐stroke had fewer conventional risk factors, smaller left atrial diameters, and higher ejection fractions, likely attributable to their younger age compared with those with AF‐stroke. IE is often a severe infection, such as bacteremia, associated with a high mortality rate. ^33^ Several laboratory tests, including white blood cell count, hemoglobin levels, platelet count, creatinine levels, and total protein, showed significant abnormalities, reflecting the severity of the infection and assisting in differentiating IE from AF. However, in multivariable analysis, only lower hemoglobin levels and higher C‐reactive protein levels were independently associated with IE. The strongest predictors of IE included smaller lesion size and a greater number of affected vascular territories.

IE‐stroke was associated with higher odds of 3‐month mortality after adjustment for baseline covariates. This increased odds likely reflects systemic factors such as persistent bacteremia, sepsis burden, and multisystemic embolism. Clinically, these findings emphasize the need for early recognition of IE in patients with embolic stroke with fever, rapid initiation of antibiotic therapy, and multidisciplinary evaluation for timely surgical intervention.

This study has several limitations. First, the retrospective nature of the study is prone to selection bias. Abdominal computed tomography was performed more frequently in the group with IE‐stroke (67.1% versus 30.7%) due to differences in fever foci workup. Although this differential imaging rate may contribute to detection bias, the substantially higher proportion of systemic embolism in the group with IE‐stroke is consistent with the known embolic propensity of infective endocarditis. Second, the selective inclusion of patients with AF‐associated stroke with fever may have led to a higher incidence of aspiration pneumonia, potentially skewing the results. This suggests that the group with AF‐stroke may not fully represent the general population of patients with AF‐stroke. Third, the inclusion of 21% of patients with AF in the group with IE‐stroke might have influenced the results. Although patients were included in the group with IE‐stroke if diagnosed with IE according to the Modified Duke criteria, the embolic risk associated with IE is believed to be higher than that from AF, suggesting that IE likely caused the ischemic strokes in these cases. Finally, baseline covariates may have demonstrated differential effects on clinical outcomes between cohorts with IE‐stroke and AF‐stroke. Subgroup analyses incorporating interaction‐term approaches revealed significant heterogeneity in the relationships between hypertension, diabetes, and prior antiplatelet therapy in relation to 3‐month mortality (Table S2). Further investigations are needed to clarify these observations. Despite its retrospective, single‐center design, the study was based on a prospective registry with standardized protocols at one of the largest tertiary centers in our country, allowing analysis of a substantial number of IE‐related stroke cases.

Conclusions

In conclusion, we found that IE‐strokes differed from AF‐strokes by typically presenting with small, multiple ischemic lesions scattered across multiple vascular territories. IE‐strokes were associated with higher rates of systemic embolism and recurrent stroke. Despite the presence of smaller lesions and fewer large vessel occlusions, IE‐strokes had a similar frequency of PH, indicating a significant tendency toward hemorrhage. These differences are likely attributable to septic emboli in IE, which are distinct from thrombi associated with AF. The number of affected vascular territories and smaller lesion sizes were independent predictors of IE. Moreover, IE‐stroke was independently associated with higher odds of 3‐month mortality, highlighting the need for early recognition and appropriate management. Brain imaging characteristics may serve as a valuable tool for the early differentiation of IE in patients with embolic strokes and fever.

Sources of Funding

This research was supported by the Digital Therapeutics Development and Clinical Validation Program, funded by the Ministry of Science and ICT (MSIT) and the National IT Industry Promotion Agency (NIPA) of Korea (Grant Number: H0601‐25‐1046).

Disclosures

The authors have nothing to disclose.

Supporting information