Highlight
In the AIS of HEARTS individual patient data meta-analysis, cardiac CT integrated into the acute ischemic stroke workup identified cardiac thrombi in 243 of 3919 patients, a diagnostic yield of 6.2%.
Among patients who underwent both tests, cardiac CT had a substantially higher diagnostic yield than transthoracic echocardiography, with an odds ratio of 7.4 (95% CI, 4.0-15.1; P<0.001).
Implementation appeared practical in the hyperacute setting: median added scan time was 6 minutes for ECG-gated protocols and 13 seconds for non-ECG-gated protocols, with a median additional radiation dose of 2.9 mSv.
Detection of cardiac thrombus identified a higher-risk subgroup, with increased 90-day mortality and worse functional outcome, although recurrent stroke rates were not significantly different.
Background
Determining stroke mechanism quickly and accurately remains one of the central clinical challenges in acute ischemic stroke. Cardioembolism is a major cause of ischemic stroke, yet the responsible source is often not identified during the acute hospital phase. Conventional transthoracic echocardiography is widely available and noninvasive, but its sensitivity for left atrial appendage thrombus and other intracardiac thrombi is limited. Transesophageal echocardiography is more sensitive for several relevant lesions, but it is semi-invasive, resource-intensive, and rarely feasible in the immediate emergency workflow.
Cardiac CT performed as an extension of acute stroke imaging has therefore attracted growing interest. In principle, it offers rapid structural assessment of the atria, ventricles, and appendages at the same time that patients already undergo vascular and brain imaging. If a cardiac thrombus can be visualized early, antithrombotic strategy, etiologic classification, and secondary prevention planning may be accelerated. However, practical concerns remain: additional imaging time, radiation burden, motion artifact, variability between ECG-gated and non-ECG-gated acquisition, and uncertainty about how often such thrombi are actually found in unselected acute stroke populations.
The AIS of HEARTS collaboration addresses this evidence gap by pooling individual patient data from four prospective cohorts to estimate the real-world diagnostic yield of cardiac CT in acute ischemic stroke and to place those findings in clinical context.
Study Design
Design and setting
This was a 1-stage individual patient data meta-analysis of four prospective observational cohorts within the AIS of HEARTS (Acute Ischemic Stroke of Heart-Related Embolic Sources Detected on Acute Cardiac CT Scans) collaboration. Patients were enrolled between May 2018 and June 2024.
Population
The analysis included patients with acute ischemic stroke who underwent either ECG-gated or non-ECG-gated cardiac CT as part of their acute evaluation. Patients with transient ischemic attack and stroke mimics were excluded. In total, 3919 patients were analyzed. The median age was 74 years (interquartile range [IQR], 63-82), 58% were male, and the median National Institutes of Health Stroke Scale (NIHSS) score was 6 (IQR, 3-12), indicating a cohort that spanned mild to moderately severe strokes.
Primary and secondary outcomes
The primary outcome was the proportion of patients with a thrombus detected on cardiac CT. Secondary outcomes included feasibility metrics such as added scan time and radiation dose, comparative diagnostic yield versus transthoracic echocardiography, and 90-day clinical outcomes.
To evaluate prognosis, the investigators used logistic regression models adjusted for clinically important covariates, including age, sex, atrial fibrillation history, ischemic heart disease, chronic heart failure, prior stroke or transient ischemic attack, anticoagulant use, prestroke modified Rankin Scale (mRS), admission NIHSS, presence of large-vessel occlusion, and intravenous thrombolysis, as appropriate for each model.
Key Findings
Diagnostic yield of cardiac CT
Cardiac CT detected a thrombus in 243 of 3919 patients, corresponding to a yield of 6.2%. That figure is clinically meaningful. In practical terms, this means that approximately 1 in 16 patients with acute ischemic stroke undergoing this imaging strategy had a directly visualized intracardiac thrombus. In a field where etiologic certainty is often elusive, that is not a marginal return.
The result also suggests that acute cardiac CT is not merely a niche test for highly selected patients. Rather, it may identify a substantial subgroup of patients with active embolic substrate during the initial stroke evaluation. Although the abstract does not fully detail thrombus location, the most clinically relevant sites in such studies are typically the left atrial appendage, left atrium, and left ventricle, each with distinct implications for rhythm evaluation, anticoagulation, and structural heart disease workup.
Comparison with transthoracic echocardiography
Among 1323 patients who underwent both cardiac CT and transthoracic echocardiography, cardiac CT showed a much higher diagnostic yield. The odds ratio was 7.4 (95% CI, 4.0-15.1; P<0.001). This is one of the most practice-relevant findings in the study.
Clinically, the comparison is not surprising. Transthoracic echocardiography has important limitations in visualizing posterior structures and the left atrial appendage, and image quality can be suboptimal in acutely ill patients. CT offers superior spatial resolution and may detect thrombi that are simply not accessible to TTE. The study therefore reinforces a distinction many clinicians already suspect: TTE is useful for global cardiac structure and function, but it is not the most sensitive test for excluding intracardiac thrombus in the acute stroke setting.
At the same time, this should not be interpreted as meaning CT replaces all echocardiography. Echocardiography remains valuable for ventricular function, valvular pathology, shunts, hemodynamics, and other structural abnormalities. The more accurate interpretation is that cardiac CT may add a high-yield anatomic thrombosis assessment early in the pathway.
Feasibility in acute workflow
Feasibility is often the deciding factor for stroke imaging innovations. Here, implementation appears realistic. The median additional scan time was 6 minutes (IQR, 5-7) for ECG-gated cardiac CT and only 13 seconds (IQR, 12-61) for non-ECG-gated cardiac CT. The median additional radiation dose was 2.9 mSv (IQR, 1.6-4.1).
These numbers matter because they frame the tradeoff against time-sensitive stroke care. For centers already performing multimodal CT imaging, an extra 13 seconds for non-ECG-gated acquisition is operationally trivial, and even a 6-minute addition for ECG-gated protocols may be acceptable if it does not delay reperfusion treatment. The radiation dose is not negligible, but it is within a range many clinicians would judge reasonable if the scan changes etiologic diagnosis or downstream treatment.
Whether ECG-gated or non-ECG-gated acquisition should be preferred remains an implementation question. ECG-gating may improve image quality and thrombus characterization but costs more time. Non-ECG-gated protocols are extremely fast and may be easier to incorporate in hyperacute practice. Future comparative analyses will be important.
Association with 90-day outcomes
Patients with thrombi had worse outcomes at 90 days. Mortality was 33% in patients with thrombus versus 15% in those without, with an adjusted odds ratio of 1.6 (95% CI, 1.1-2.3). Functional outcome was also worse: the median mRS was 3 versus 2, with an adjusted odds ratio of 1.6 (95% CI, 1.2-2.0).
These findings suggest that a visible cardiac thrombus is not only a diagnostic marker but also a prognostic marker. It may identify patients with more advanced atrial cardiopathy, ventricular dysfunction, systemic thrombogenicity, or more severe embolic burden. The association persisted after multivariable adjustment, which strengthens the inference that thrombus presence captures clinically important risk not fully explained by baseline characteristics alone.
Interestingly, recurrent stroke rates were similar: 5% in patients with thrombus and 4% in those without, adjusted odds ratio 1.4 (95% CI, 0.7-2.5). This confidence interval crosses unity, so the study does not support a clear increase in early recurrent stroke. Several explanations are plausible. First, thrombus detection may have prompted treatment escalation, reducing recurrence risk. Second, mortality may have acted as a competing event. Third, recurrent stroke may simply be too infrequent for this comparison to be precise despite the overall large sample.
Clinical Interpretation
The main contribution of this study is not just that cardiac CT can find thrombi, but that it can do so at a meaningful rate during the acute stroke workup without imposing a prohibitive workflow penalty. For stroke teams, this shifts cardiac CT from an interesting adjunct to a potentially strategic component of etiologic evaluation.
The study also sharpens the conversation around test selection. If the clinical question is specifically, “Is there a cardiac thrombus that could explain this stroke?” then cardiac CT appears substantially more sensitive than transthoracic echocardiography in this setting. That matters for patients with embolic-appearing infarcts, newly detected atrial fibrillation, large-vessel occlusion without clear atherosclerotic source, or suspected left ventricular thrombus after myocardial injury.
From a systems perspective, this approach is especially attractive in comprehensive stroke centers where CT-based triage is already dominant. Integrating a cardiac phase into a single-session protocol may shorten diagnostic latency compared with arranging delayed inpatient echocardiography. Earlier source identification could influence anticoagulation timing, rhythm monitoring strategies, and discharge planning.
Strengths and Limitations
Strengths
This analysis has several strengths. It used individual patient data rather than aggregate study-level data, allowing more consistent outcome definitions and adjusted prognostic modeling. The cohorts were prospective, the sample size was large, and the study reflects contemporary practice across multiple sites over several years. The inclusion of both feasibility metrics and clinical outcomes makes the findings more actionable than a purely diagnostic study.
Limitations
Important limitations remain. The underlying cohorts were observational, so the study cannot determine whether adding cardiac CT improves patient outcomes compared with standard care. It also cannot establish the best management strategy after thrombus detection. The abstract does not provide full details on reference standards for thrombus verification, location-specific performance, treatment changes triggered by CT, or intersite heterogeneity in protocol quality and interpretation.
The comparison with transthoracic echocardiography should also be interpreted as a comparison of diagnostic yield rather than a formal sensitivity-specificity study against a universal gold standard. Some CT-detected filling defects may represent slow flow rather than definite thrombus, particularly in the left atrial appendage if delayed imaging protocols are not standardized. Conversely, some thrombi may still be missed on CT, especially with non-ECG-gated acquisition or suboptimal contrast timing.
Generalizability may differ by center resources, scanner quality, and local stroke workflow. Hospitals without established acute multimodal CT pathways may find implementation more difficult than the study’s headline feasibility estimates imply.
Implications for Practice and Research
For clinicians, the immediate takeaway is that acute cardiac CT deserves consideration when a center has the technical capability to incorporate it without delaying reperfusion therapy. It is most compelling in patients with presumed embolic stroke, uncertain mechanism, or high suspicion for a cardiac source not likely to be resolved by TTE alone.
For researchers, the next questions are now clear. Randomized or stepped-wedge implementation studies are needed to determine whether routine acute cardiac CT changes management, reduces time to etiologic diagnosis, and improves hard outcomes. Studies should also clarify which acquisition strategy is optimal, whether delayed phase imaging improves specificity, and which patient subgroups derive the greatest value.
Another important frontier is treatment consequence. Detecting thrombus has intuitive therapeutic implications, but optimal anticoagulation timing after acute ischemic stroke remains nuanced and depends on infarct size, hemorrhagic transformation risk, and thrombus location. Imaging only becomes practice-changing when linked to a validated management pathway.
Conclusion
The AIS of HEARTS meta-analysis provides strong contemporary evidence that cardiac CT can be feasibly added to acute ischemic stroke imaging and will detect cardiac thrombi in about 6% of patients. Its diagnostic yield exceeds that of transthoracic echocardiography, and thrombus detection identifies a subgroup with significantly higher mortality and worse functional outcome at 90 days. While the study does not prove that routine cardiac CT improves outcomes, it meaningfully advances the case for cardiac CT as an early etiologic tool in modern stroke care.
Funding and ClinicalTrials.gov
ClinicalTrials.gov registration: NCT07165093.
Funding information was not reported in the provided abstract and citation details.
References
Nio SS, Green DS, Berry-Noronha A, Rinkel LA, Beemsterboer CFP, Hasnain MG, Planken RN, van Randen A, Al-Hadethi S, Hilt MGM, Ruhe DSA, Bouma BJ, Boekholdt SM, Chew BLA, Fridgant J, Mena-Romo L, Kim DH, Guo S, Alamri Y, Winders J, Senadeera SC, Lim AT, Fink JN, Leung M, Cordato D, Parsons MW, Spratt NJ, Garcia-Esperon C, Wu TY, Coutinho JM, AIS of HEARTS Collaborators. Diagnostic Yield of Cardiac CT to Detect Cardiac Thrombi in Patients With Acute Ischemic Stroke (AIS of HEARTS). Stroke. 2026-06-02. PMID: 42227105. URL: https://pubmed.ncbi.nlm.nih.gov/42227105/
