Cardiac CT vs TTE
Diagnostic Yield of ECG-Gated Cardiac CT in the Acute Phase of Ischemic Stroke vs Transthoracic Echocardiography
Clinical Question
Does cardiac computed tomography (CT), acquired during the initial stroke imaging protocol, have a higher yield for detecting high-risk cardioaortic sources of embolism than transthoracic echocardiography (TTE) in patients with ischemic stroke?
Bottom Line
Cardiac CT acquired during the initial hyperacute stroke imaging protocol has a significantly higher diagnostic yield for detecting high-risk cardioaortic sources of embolism than transthoracic echocardiography (TTE) in patients with acute ischemic stroke (11.4% vs 4.9%, OR 5.60), driven largely by cardiac thrombi (7.1% vs 0.6%). Cardiac CT may be considered as a first-line screening method for cardioembolism.
Major Points
- 452 adults with acute ischemic stroke were included (774 screened; 228 excluded for non-ischemic diagnosis, 94 for no consent); 350 (77.4%) also underwent TTE. Median age 72 (IQR 62-81); 59.3% male; median NIHSS 5.
- Primary outcome (within-patient comparison, n=350): high-risk cardioaortic source of embolism detected in 40 (11.4%) on cardiac CT vs 17 (4.9%) on TTE (OR 5.60, 95% CI 2.28-16.33) by 2-sided McNemar test.
- Cardiac thrombus was the most common finding: 25 (7.1%) on CT vs 2 (0.6%) on TTE; left atrial appendage thrombus 19 (5.4%) on CT vs 0 on TTE.
- In the full study population (n=452), high-risk source detected in 55 (12.2%) on CT vs 17 (3.8%) on TTE (OR 5.99, 95% CI 2.34-15.34).
- Among 175 patients with cryptogenic stroke, cardiac CT identified a cause in 11 (6.3%); among 94 ESUS patients, high-risk source in 6 (6.4%).
- Cardiac CT was feasible: good/excellent image quality in 73.2%, median additional scan time 6 minutes (IQR 5-7), effective radiation dose 2.0 mSv (IQR 1.3-3.4).
- TTE detected sources missed by CT in 8 patients (rheumatic valve disease 4, papillary fibroelastoma 2, recent MI 1, endocarditis 1), changing management in 2.
- Interrater agreement for high-risk cardioaortic sources on cardiac CT was substantial (Cohen κ 0.78).
Design
Study Type: Prospective, single-center, observational cohort study
Randomization:
Blinding: Cardiac CT and echocardiography were independently re-evaluated for study purposes by a cardiac radiologist (R.N.P. or A.v.R.) and a cardiologist (S.M.B. or B.J.B.), blinded to each other's results.
Enrollment Period: May 2018 to November 2020
Follow-up Duration: Structured telephone interviews at 3 months and 2 years (2-year follow-up ongoing at publication); primary analysis focused on diagnostic yield at presentation.
Centers: 1
Countries: Netherlands
Sample Size: 452
Analysis: Primary outcome (within-patient comparison, n=350) analyzed with 2-sided McNemar test with binomial enumeration (mid-p). Full-population comparison (n=452) used multiple imputation by chained equations (5 imputations) and conditional logistic regression. Interobserver variability by Cohen κ. Baseline characteristics compared with 2-sided independent t-test, χ2, or Fisher exact test. Significance level 0.05. Analyses in R version 4.0.3.
Inclusion Criteria
- Consecutive adult patients with acute ischemic stroke (diagnosis established by neurologist based on clinical and imaging data; suspected cases screened and non-ischemic diagnoses excluded).
- Potentially eligible for reperfusion therapy (IV thrombolysis or endovascular treatment) at time of admission.
- Acute onset neurologic symptoms developed less than 24 hours prior.
- Underwent ECG-gated cardiac CT during the initial stroke imaging protocol.
- Written informed consent obtained from patient or legal representative.
Exclusion Criteria
- Transient ischemic attack (TIA).
- Diagnosis other than ischemic stroke established after workup (228 patients excluded post-screening).
- Unable to obtain written informed consent (94 patients, including declined participation, impaired capacity, or verbal-only consent).
Arms
| Field | Cardiac CT | Control |
|---|---|---|
| Intervention | Prospective ECG-gated sequential cardiac CT (end-diastole) added to the hyperacute stroke imaging protocol, performed after non-contrast brain CT, CT perfusion, and non-gated CTA of the aortic arch and cervical/intracranial arteries, using a third-generation dual-source scanner (Somatom Force). | Routine TTE per Dutch and European stroke guidelines (standard of care), performed as soon as possible after admission; patients <60 years received contrast TTE with agitated saline for PFO detection. |
| Duration | Acute stroke evaluation; median acquisition time 4 seconds (IQR 3-4); median 6 minutes (IQR 5-7) additional time from end of cervical CTA to start of cardiac CT. | Acute stroke workup; median interval between cardiac CT and TTE was 1 day (IQR 0-13); 99/350 TTEs performed in outpatient setting. |
Outcomes
| Outcome | Type | Control | Intervention | HR / OR / RR | P-value |
|---|---|---|---|---|---|
| Proportion of patients with a predefined high-risk cardioaortic source of embolism on cardiac CT compared with TTE in patients who underwent both investigations (n=350). High-risk sources: cardiac thrombus, prosthetic valve abnormalities, endocarditis, atrial myxoma, papillary fibroelastoma, recent MI, rheumatic valvular disease, >4 mm ulcerated noncalcified aortic arch plaque proximal to affected territory, Stanford Type A aortic dissection. | Primary | 17/350 (4.9%) | 40/350 (11.4%) | OR 5.60 (95% CI 2.28-16.33) | |
| Any high-risk cardioaortic source of embolism in full study population (N=452) | Secondary | 17 (3.8%) | 55 (12.2%) | 5.99 (95% CI 2.34-15.34) | |
| Cardiac source of embolism (any) — CT vs TTE (n=350) | Secondary | 13 (3.7%) | 37 (10.6%) | ||
| Cardiac thrombus (any location) — CT vs TTE (n=350) | Secondary | 2 (0.6%) | 25 (7.1%) | ||
| Left atrial appendage thrombus — CT vs TTE (n=350) | Secondary | 0 (0%) | 19 (5.4%) | ||
| Left atrial thrombus — CT vs TTE (n=350) | Secondary | 0 (0%) | 1 (0.3%) | ||
| Left ventricular thrombus — CT vs TTE (n=350) | Secondary | 2 (0.6%) | 6 (1.7%) | ||
| Signs of endocarditis — CT vs TTE (n=350) | Secondary | 3 (0.9%) | 6 (1.7%) | ||
| Prosthetic valve pannus or thrombus — CT vs TTE (n=350) | Secondary | 1 (0.3%) | 3 (0.9%) | ||
| Atrial myxoma — CT vs TTE (n=350) | Secondary | 1 (0.3%) | 1 (0.3%) | ||
| Papillary fibroelastoma — CT vs TTE (n=350) | Secondary | 2 (0.6%) | 0 (0%) | ||
| Recent myocardial infarction (<4 wk) — CT vs TTE (n=350) | Secondary | 1 (0.3%) | 2 (0.6%) | ||
| Signs of rheumatic valvular disease (mitral stenosis) — CT vs TTE (n=350) | Secondary | 4 (1.1%) | 0 (0%) | ||
| Aortic source of embolism (any) — CT vs TTE (n=350) | Secondary | 4 (1.1%) | 5 (1.4%) | ||
| Stanford Type A aortic dissection — CT vs TTE (n=350) | Secondary | 4 (1.1%) | 4 (1.1%) | ||
| >4 mm ulcerated noncalcified aortic arch plaque — CT vs TTE (n=350) | Secondary | NA (not assessable by TTE) | 1 (0.3%) | ||
| Diagnostic yield of cardiac CT in cryptogenic stroke (n=175) | Secondary | 11 (6.3%) — 8 cardiac thrombus, 2 aortic thrombus, 1 endocarditis | |||
| Diagnostic yield of cardiac CT in ESUS patients (n=94) | Secondary | 6 (6.4%) | |||
| Interrater agreement for high-risk source detection on cardiac CT (Cohen κ) | Secondary | 0.78 (substantial) | |||
| 90-day recurrent ischemic stroke (overall) | Secondary | 22 (5.1%); 7.4% in patients with high-risk source on CT vs 4.8% without | |||
| Not systematically reported | Adverse | Observational diagnostic imaging study — no formal adverse event framework. Additional radiation dose from cardiac CT: median dose-length product 146 (IQR 95-241), effective dose 2.0 mSv (IQR 1.3-3.4). |
Subgroup Analysis
Sex-stratified analysis (prespecified secondary outcome): no significant difference in proportion of men vs women with high-risk cardioaortic source on CT (10.8% vs 14.1%, p=0.29). Cryptogenic stroke subgroup (n=175): CT identified cause in 11 (6.3%). ESUS subgroup (n=94): CT detected high-risk source in 6 (6.4%).
Criticisms
- Single-center design (Amsterdam UMC) limits generalizability.
- Observational cohort — no randomization; 22.6% of included patients did not undergo TTE (mostly due to early death or outpatient burden), potentially biasing the comparison.
- Comparison was against TTE rather than TEE, which has higher diagnostic yield for LAA thrombus and is considered the gold standard.
- Brain MRI was performed in only a minority of patients (9.7%), which may have inflated the proportion classified as cryptogenic stroke.
- Delayed-phase cardiac CT scanning (which improves specificity for distinguishing LAA thrombus from slow-flow) was not performed, though prior contrast administration for CTA/perfusion partially mimicked delayed imaging.
- Study was not primarily designed to assess the effect of CT-detected findings on clinical management or long-term outcomes.
- Prospective ECG-gating (single cardiac phase) precludes calculation of LVEF and dynamic assessment of valves, and PFO sensitivity in the acute phase was not properly evaluated.
Subgroup Analysis
Prespecified secondary outcome included sex-stratified analysis (men 10.8% vs women 14.1%, p=0.29 for high-risk source on CT). Cryptogenic stroke (n=175) and ESUS (n=94) results are reported descriptively (cause identified on CT in 11 [6.3%] and high-risk source in 6 [6.4%], respectively) but were not labeled as prespecified subgroup analyses. Interrater variability assessed in a sample enriched for cardioaortic abnormalities (Cohen κ 0.78).
Funding
V. Guglielmi received research grants from the Royal Netherlands Academy of Arts and Sciences (Van Leersum Grant), Foundation De Drie Lichten, Remmert Adriaan Laan Foundation, and AMC Young Talent Fund (all non-profit research foundations). Institutional research grants disclosed for co-authors: CVON/Dutch Heart Foundation, European Commission, TWIN Foundation Dutch Health Evaluation program, and Stryker (C.B.L.M. Majoie); Medtronic (J.M. Coutinho). No industry funding for the study itself.
Based on: Cardiac CT vs TTE (Neurology, 2022)
Authors: Leon A. Rinkel, MD, Valeria Guglielmi, ..., and Jonathan M. Coutinho
Citation: Neurology® 2022;99:e1456-e1464. doi:10.1212/WNL.0000000000200995
Content summarized and formatted by NeuroTrials.ai.