← Back

Cardiac CT in LVO Stroke

Cardiac CT in Large Vessel Occlusion Stroke for the Evaluation of Non-Thrombotic and Non-Atrial-Fibrillation-Related Embolic Causes

Share Share Copied! Compare

Year of Publication: 2025

Authors: Karim Mostafa, Cosima Wünsche, Sarah Krutmann, ..., Patrick Langguth

Journal: Neurology International

Citation: Neurol Int. 2025;17:25.

Link: https://doi.org/10.3390/neurolint17020025

PDF: https://mdpi-res.com/d_attachment/neurol...sion=1738922506

Clinical Question

In patients with acute LVO stroke without atrial fibrillation or intracardiac thrombus, which cardiac CT findings are independently associated with a cardioembolic stroke aetiology?

Bottom Line

In 121 LVO stroke patients without AF or intracardiac thrombus, five cardiac CT findings were independently associated with cardioembolic aetiology: LV dilatation (AOR 32.4; P=0.004), visible interatrial R-to-L shunt (AOR 30.8; P=0.006), aortic/mitral valve implants (AOR 24.5; P=0.009), aortic arch atheroma grade >II (AOR 6.9; P=0.015), and post-ischemic myocardial scar (AOR 6.3; P=0.032). Combined model AUC=0.83.

        Major Points
        Cardioembolic cause in 70.2% of all LVO patients (221/315); AF+thrombus explained 86.9%.
After excluding AF/thrombus, only 24% (29/121) cardioembolic — vs 47.1% ESUS.
LV dilatation strongest predictor: AOR 32.4 (95% CI 3.0-349; P=0.004).
Visible interatrial R→L shunt: AOR 30.8 (95% CI 2.7-341.3; P=0.006) — present in 3.3%.
Valve implants (aortic/mitral): AOR 24.5 (95% CI 2.2-270.9; P=0.009) — present in 4.1%.
Aortic arch atheroma grade >II (≥4mm): AOR 6.9 (95% CI 1.5-32.8; P=0.015) — present in 10.7%.
Post-ischemic myocardial scar: AOR 6.3 (95% CI 1.2-34.1; P=0.032) — present in 9.1%.
Combined 5-factor model AUC=0.83.
ESUS was largest aetiology (47.1%) in the analytic cohort — highlighting need for better detection tools.
Cardiac CT additional radiation: median DLP only 265 mGy*cm — low relative to benefit.

      

Design

Study Type: Retrospective observational cohort study

Randomization:

Blinding: Not applicable (retrospective imaging analysis)

Enrollment Period: 2018-2024

Follow-up Duration: Inpatient stroke unit evaluation

Centers: 1

Countries: Germany

Sample Size: 121

Analysis: Multivariate logistic regression; ROC analysis. Full cohort 315, analytic cohort 121 (no AF, no thrombus).

Inclusion Criteria

Intracranial large vessel occlusion confirmed on CT angiography.
Cardiac CT available from acute stroke setting.
Absence of intracardiac thrombus on cardiac CT.
Absence of atrial fibrillation (history, chart, 24h ECG).

Exclusion Criteria

Known or newly diagnosed atrial fibrillation.
Intracardiac thrombus on cardiac CT.
No available cardiac CT from acute setting.
No confirmed LVO on CTA.

Baseline Characteristics

Analytic Cohort (N=121, no AF/thrombus):

Female sex: 64 (52.9%)
Age (mean±SD): 69.4±16.1
Hypertension: 66 (54.5%)
Diabetes: 18 (14.9%)
Smoker: 28 (23.1%)
Hyperlipidemia: 25 (20.7%)
CHF: 9 (7.4%)
TTE performed: 82 (67.8%)
TEE performed: 26 (21.5%)
TOAST Cardioembolic: 29 (24.0%)
TOAST Large artery: 28 (23.1%)
TOAST ESUS: 57 (47.1%)
Vessel — MCA: majority
Vessel — Carotid-T: subset

Arms

Field	Observational cohort
Intervention	Retrospective analysis of cardiac CT findings in LVO stroke patients without AF/thrombus. TOAST classification as outcome.
Duration	Inpatient

Outcomes

Outcome	Type	Control	Intervention	P-value
Independent cardiac CT predictors of cardioembolic aetiology (multivariate logistic regression)	Primary	Non-cardioembolic (n=92)	Cardioembolic (n=29)	See individual predictors
LV dilatation → cardioembolic \| 95% CI: 3.0-349	Secondary	—	AOR 32.4	0.004
Visible interatrial R→L shunt → cardioembolic \| 95% CI: 2.7-341.3	Secondary	—	AOR 30.8	0.006
Aortic/mitral valve implant → cardioembolic \| 95% CI: 2.2-270.9	Secondary	—	AOR 24.5	0.009
Aortic arch atheroma grade >II → cardioembolic \| 95% CI: 1.5-32.8	Secondary	—	AOR 6.9	0.015
Post-ischemic myocardial scar → cardioembolic \| 95% CI: 1.2-34.1	Secondary	—	AOR 6.3	0.032
Observational imaging study	Adverse		Observational study of cardiac CT for embolic source detection - no AE data

Subgroup Analysis

No formal subgroup analyses. Primary regression performed on n=121 analytic subcohort.

Criticisms

Very wide confidence intervals (AOR 3.0-349 for LV dilatation) — small events, likely overfitting.
Only 29 cardioembolic events with 5 predictor variables — violates rule of 10 events/variable.
Single-center retrospective design — cannot establish causality.
TOAST classification by interdisciplinary consensus, not fully blinded adjudication.
No comparison with TEE as gold standard for all patients (TEE only in 21.5%).
Selection bias: only LVO patients without AF/thrombus — small minority of all stroke.
No anticoagulation/antiplatelet data considered.

Funding

No external funding received.

Based on: Cardiac CT in LVO Stroke (Neurology International, 2025)