← Back

ENCLOSE

Detection of Cardioembolic Sources With Nongated Cardiac Computed Tomography Angiography in Acute Stroke: Results From the ENCLOSE Study

Share Share Copied! Compare

Year of Publication: 2023

Authors: Frans Kauw, Birgitta K. Velthuis, Richard A.P. Takx, ..., Jan W. Dankbaar

Journal: Stroke

Citation: Stroke. 2023;54:821–830. doi:10.1161/STROKEAHA.122.041018

Link: https://www.ahajournals.org/doi/10.1161/STROKEAHA.122.041018

PDF: https://www.ahajournals.org/doi/reader/1...EAHA.122.041018

Clinical Question

Can nongated cardiac computed tomography angiography (CTA), acquired during acute stroke imaging, detect cardioembolic sources such as cardiac thrombus and improve diagnostic confidence?

Study Overview

Objective

To evaluate the diagnostic yield of admission nongated cardiac CTA, including spectral reconstructions, for detecting cardioembolic sources in patients with acute ischemic stroke or TIA.

Study Summary

Nongated cardiac CTA extended from standard stroke imaging identified cardiac thrombi in 12% of patients and significantly improved diagnostic certainty of cardioembolic stroke.

Intervention

Admission nongated head-to-heart CTA extended from the routine stroke protocol; spectral reconstructions (monoenergetic 40 keV, iodine maps, Z-effective) were acquired in 271 of 370 patients (73%).

Patients per Arm

370 patients with assessable cardiac CTA (44 thrombus-positive, 309 thrombus-negative); spectral reconstructions in 271 (73%).

Bottom Line

Admission nongated cardiac CTA identified cardiac thrombi in 12% of patients and increased diagnostic certainty of a cardioembolic stroke mechanism; iodine spectral maps (available in 73% of scans) provided additional diagnostic value, especially for left atrial appendage thrombus.

        Major Points
        ENCLOSE demonstrated that nongated cardiac CTA — performed as a simple EXTENSION of routine acute stroke CTA (no additional contrast, no ECG gating) — can detect cardiac thrombus in 12% of stroke patients.
Left atrial appendage (LAA) thrombus was found in 9%, left ventricular (LV) thrombus in 4% — these are immediately actionable findings that mandate anticoagulation.
Spectral iodine mapping (dual-energy CT technique, performed in 271/370 patients) significantly improved diagnostic certainty for LAA thrombus vs standard CTA — the iodine map distinguishes slow-flow (stasis) from true thrombus.
In the 44 thrombus-positive patients, expert-panel likelihood of a cardioembolic source rose from 0% 'certainly yes' before cardiac CTA review to 68% 'certainly yes' after review; combined 'probably or certainly yes' rose from 55% to 86%.
Patients with cardiac thrombus had higher NIHSS (median 10 vs 4), more frequent AF (43% vs 13%), and higher rates of endovascular treatment (43% vs 20%) — consistent with cardioembolic strokes being more severe.
Clinical implication: extending the CTA scan field to cover the heart adds <5 seconds of scan time and no additional iodine contrast, with an acceptable ~2 mSv increase in radiation dose.
Challenges the traditional stroke workup paradigm where cardiac imaging (TTE/TEE) is done days later — cardiac CTA provides thrombus detection at the time of acute stroke imaging.
Limitations: nongated CT has lower sensitivity than ECG-gated cardiac CT for small LAA thrombi. Some findings may be false positives (slow-flow artifact mimicking thrombus).
Single-center Dutch study (n=370) — needs multicenter validation; no formal reference-standard comparison (paper notes TEE/TTE do not outperform CTA).
Aligns with the growing 'one-stop-shop' CT approach to acute stroke workup: head CT → CTA brain + neck + heart → CT perfusion, all in a single session.

      

Design

Study Type: Prospective observational cohort study

Randomization:

Blinding: Imaging observers blinded to clinical data; expert panel used staged unblinding across two phases

Enrollment Period: June 2017 to March 2022

Follow-up Duration: 90 days for selected clinical outcomes

Centers: 1

Countries: Netherlands

Sample Size: 370

Analysis: Descriptive statistics; parametric and nonparametric tests; χ² for secondary/tertiary outcomes; R v3.5.1

Inclusion Criteria

Age ≥18 years
Clinical diagnosis of TIA or acute ischemic stroke
Admission CT imaging within 9 hours of symptom onset or last seen well
Assessable nongated cardiac CTA

Exclusion Criteria

Patients without an assessable cardiac CTA were excluded (n=13 no cardiac CTA performed; n=6 very poor contrast).

Baseline Characteristics

Characteristic	Thrombus (n=44)	No thrombus (n=309)
Age (mean±SD)	71±14	67±14
Male (%)	28 (64%)	186 (60%)
Admission NIHSS (median, Q1–Q3)	10 (3–17)	4 (1–10)
History of AF	19 (43%)	39 (13%)
History of myocardial infarction	11 (25%)	33 (11%)
VKA use	12 (27%)	17 (6%)
DOAC use	4 (9%)	32 (10%)
Endovascular treatment	19 (43%)	61 (20%)
TOAST cardioembolism	27 (61%)	57 (18%)

Arms

Field	Cardiac CTA cohort (single arm)
Intervention	Admission nongated head-to-heart CTA extended from the routine stroke protocol; spectral reconstructions (iodine maps, monoenergetic 40 keV, Z-effective) were acquired in 271 of 370 patients (73%).
Duration	Admission imaging (single acquisition)

Outcomes

Outcome	Type	Control	Intervention	P-value
Presence of cardiac thrombus on admission nongated cardiac CTA (single-arm diagnostic yield; no comparator, so no NNT/ARR).	Primary		44/370 (12%) had cardiac thrombus: 35 (9%) in LAA, 14 (4%) in LV (5 in both).
Diagnostic usefulness of spectral iodine maps for LAA thrombus, thrombus-positive vs thrombus-negative patients (not a standard-CTA control comparison).	Secondary	Patients without cardiac thrombus: iodine map rated useful in 17%.	Patients with cardiac thrombus: iodine map rated useful in 50%.	<0.001
Expert-panel likelihood of a cardioembolic source in the 44 thrombus-positive patients, before vs after cardiac CTA review (staged unblinding).	Secondary	Phase 1 (before CTA): Certainly yes 0/44 (0%); Probably yes 24/44 (55%); Possibly yes 20/44 (45%); combined probably/certainly yes 24/44 (55%).	Phase 2 (after CTA): Certainly yes 30/44 (68%); Probably yes 8/44 (18%); Possibly yes 6/44 (14%); combined probably/certainly yes 38/44 (86%).	<0.001
Procedural burden (imaging study — no formal AE dataset)	Adverse		Extending stroke CTA to cover the heart added <5 s of extra scan time, required no additional iodine contrast, and produced a ~2 mSv increase in radiation dose (paper describes this as an acceptable increase).

Criticisms

Single-center study (University Medical Center Utrecht) — imaging protocols, scanner technology (spectral CT), and radiologist expertise may not be available at all stroke centers.
No reference standard for cardiac thrombus detection — the paper explicitly notes TEE and TTE 'do not outperform CTA,' so sensitivity/specificity could not be analyzed; false-positive/negative rates of nongated cardiac CTA remain unquantified.
Nongated CT cannot reliably measure left atrial enlargement (a known stroke risk factor) because cardiac phase differs between scans; PFO, valve abnormalities, cardiomyopathy, and LV aneurysm WERE assessed on the cardiac CTA in this study.
Observational design with no clinical outcomes endpoint — detecting thrombus is informative, but the study did not demonstrate that cardiac CTA-guided management improved stroke outcomes.
Only the 44 thrombus-positive patients were presented to the expert panel — this could bias decision-making, though staged unblinding was used to mitigate incorporation bias.
Small sample size (n=370) with only 44 thrombus-positive patients — subgroup analyses are underpowered and associations may be chance findings.
Selection bias: only patients with assessable cardiac CTA were included, and many patients could not be enrolled due to COVID-19 restrictions and early discharge (though the authors argue selection bias risk is low).
Spectral CT (dual-energy) technology was available in only 271/370 (73%) scans and is not universally available — standard single-energy CTA may have lower sensitivity for LAA thrombus.
No formal inter-rater reliability metric reported for cardiac CTA thrombus detection across all cases (the first 24 scans were used to align observers).

Funding

Dutch Heart Foundation and Netherlands Organization for Scientific Research (grant #14732)

Based on: ENCLOSE (Stroke, 2023)