Accurate detection of left ventricular (LV) thrombus affects clinical outcomes and therapeutic management, as thrombus provides a substrate for thromboembolic events and a rationale for anticoagulation. Relative risk (RR) for thrombus development is highest among subjects with systolic heart failure or myocardial infarction (MI), reflecting a link between impaired LV blood stasis, pro-coagulant mediators, and thrombosis. As the prevalence of heart failure and coronary artery disease (CAD) continues to increase, the clinical importance of accurate diagnostic imaging for thrombus is heightened.
Non-contrast echocardiography (echo) is widely used to detect thrombus based on its anatomical appearance. This imaging approach is effective when thrombus is large in size or protuberant (intracavitary) in shape, but can be challenging when thrombus is small or flat (mural). Recent studies have reported that up to two-thirds of LV thrombi can be missed by routine non-contrast echo, with mural or small thrombi least likely to be detected.1–4 These limitations have spurred the use of contrast-enhanced imaging for improved thrombus detection. Two major imaging approaches have been used for this purpose: contrast can be used to opacify the LV cavity and thereby facilitate detection of thrombus based on anatomical appearance; alternatively, contrast can be used to identify thrombus based on tissue characteristics. Each of these contrast-enhanced imaging approaches has been shown to provide incremental value over non-contrast echo for detection of LV thrombus.
This article reviews the role of cardiac imaging for LV thrombus, with a focus on recent improvements in thrombus detection as provided by contrast-enhanced imaging.
Definition of Left Ventricular Thrombus
Thrombus has traditionally been identified based on its anatomical appearance (see Figure 1A). Echo studies have generally defined thrombus as a mass within the LV cavity with margins distinct from ventricular endocardium and distinguishable from papillary muscles, chordae, trabeculations, or technical artifacts.5,6 Additional features, such as pattern of mobility,7 size,8 and associated ventricular wall motion abnormalities,9–13 have each been used as adjuvant diagnostic criteria for thrombus. However, definition of thrombus based on anatomical appearance or functional criteria has several potential pitfalls. First, technical artifacts, such as ‘near field’ artifacts on echo imaging, can mimic thrombus and obscure assessment of the LV cavity. Second, other LV masses such as neoplasms can appear similar to thrombus and be difficult to discriminate based on anatomical appearance alone. Third, thrombus may vary in size and shape and therefore be difficult to identify using a uniform anatomical criterion. For example, intracavitary thrombus typically appears as a distinct mass whereas mural thrombus is contiguous with surrounding myocardium. However, like intracavitary thrombus, mural thrombus can affect clinical outcomes; in prior studies, up to 40% of embolic events have occurred in patients with non-protuberant or immobile thrombus,14 emphasizing the importance of thrombus detection irrespective of shape or mobility.
Tissue Characterization Criteria
While thrombus can vary in shape or size, it is intrinsically characterized by its avascular tissue properties. As thrombus is composed of coagulated blood, it has no inherent vascular supply. This fundamental characteristic—avascularity—is a distinguishing feature that can be used to discriminate thrombus from both surrounding myocardium and other cardiac masses that can mimic thrombus in anatomical appearance. For example, while neoplasm and thrombus can be similar in shape, the former is reliant upon vascular supply whereas the latter is not. Recent advances in cardiac imaging have enabled assessment of vascular supply, allowing thrombus to be identified by avascular tissue characteristics resulting from absence of contrast uptake. Using this criterion, thrombus has been defined as an LV mass with tissue characteristics consistent with avascular tissue, identifiable as a low-signal- intensity mass surrounded by high-signal-intensity structures such as cavity blood and/or surrounding myocardium.2–4Figure 1 provides a representative example of tissue characterization imaging for LV thrombus (B), with a side-by-side comparison with thrombus assessment using anatomical imaging (A).
Intravenous contrast agents have expanded imaging capabilities for thrombus detection, yielding improvements in both anatomical and tissue characterization approaches. The benefits of contrast-enhanced imaging for thrombus have been demonstrated for multiple cardiac imaging modalities, including echo, magnetic resonance imaging (MRI), and computed tomography (CT).2–4,15–17
Echo has typically used contrast to facilitate anatomically based assessment of thrombus. For this purpose, sonographic contrast agents (microspheres) are administered as a bolus or continuous intravenous infusion. Thrombus detection is facilitated as a consequence of improved LV endocardial border definition and LV cavity visualization, with thrombus defined based on standard anatomic criteria (see ‘Definition of Left Ventricular Thrombus—Anatomical Criteria’, above). Figure 2 provides a typical example of improved detection of thrombus provided by contrast-enhanced cavity opacification.
Multiple studies have demonstrated that contrast echo markedly improves thrombus detection in patients with sub-optimal-quality non-contrast echoes. This concept was demonstrated by Mansecal et al., who performed non-contrast and contrast echo in a cohort of 50 consecutive patients who had sustained anterior MI.18 Both tests were performed seven days following MI and read independently. Contrast echo significantly improved LV cavity assessment, enabling full apical visualization in 100% of patients compared with 89% with non-contrast echo (p<0.0001). Improved cavity visualization facilitated thrombus detection: contrast echo detected LV thrombus in 12% of patients (n=6), only half of whom had thrombus detected by non-contrast echo. Similarly, in a study by Thanigaraj et al., use of sonographic contrast yielded a 90% reduction in the number of echoes interpreted as non-diagnostic for LV thrombus.19 The clinical utility of contrast echo was also demonstrated by Kurt et al., who studied 632 consecutive patients with technically difficult non-contrast echoes.20 Compared with non-contrast echo, contrast echo reduced the number of uninterpretable (11.7 to 0.3%) and technically difficult studies (86.7 to 9.8%; both p<0.0001). Contrast echo also yielded a greater than two-fold increase in patients with definite thrombus (eight versus three), while markedly reducing the number of patients with suspected but indefinite thrombus (35 versus one; p<0.0001).
The diagnostic benefits of contrast echo have been linked to improvements in clinical care. For example, among the cohort studied by Kurt et al.,20 contrast echo use resulted in avoidance of additional diagnostic procedures in 32.8% of patients and alteration in therapeutic management in 10.4%, with a total impact (procedures avoided, change in therapy) in 35.6% of patients. In this study, cost–benefit analysis demonstrated significant cost savings ($122/patient) resulting from echo contrast use.
Consensus guidelines currently recommend that echo contrast be utilized when non-enhanced images are sub-optimal for definitive diagnosis.21 Recent product safety considerations have prompted the US Food and Drug Administration (FDA) to issue a boxed label warning that includes revised contraindications to the use of perflutren-containing echo contrast (Definity).22 Use of this contrast agent is prohibited for patients with known or suspected right-to-left, bi-directional, or transient right-to-left cardiac shunts, as well as for patients with hypersensitivity to perflutren contrast agents. Additionally, for patients with pulmonary hypertension or unstable cardiopulmonary conditions, manufacturer guidelines recommend that monitoring of vital signs, electrocardiography (ECG), and cutaneous oxygen saturation be performed during and for at least 30 minutes following administration, with resuscitation and trained personnel readily available.
While it is important to recognize that rare but serious cardiopulmonary reactions (including fatalities) have occurred during or following administration of perflutren-containing microsphere contrast, multiple outcomes studies have demonstrated that adverse events are very uncommon. In a study of 18,671 hospitalized patients who underwent clinically requested echo, Kusnetzky et al. reported no difference in acute mortality (24 hours post-echo) between patients who received echo contrast (n=6,196) and those who underwent non-contrast echo alone (n=12,475).23 Both the contrast and non-contrast groups demonstrated low mortality (0.42 versus 0.37%; p=0.60), despite increased clinical acuity and comorbid conditions among patients who underwent contrast echo.
In another study of 42,408 patients who underwent contrast echo, Dolan et al. reported no difference in immediate (one-hour) or 30-day risk for death or MI among patients who received echo contrast compared with a matched cohort who did not receive echo contrast.24 Among a subgroup of 23,659 patients who underwent rest (non-stress) contrast echo, the combined 24-hour event rate was 0.02% (4/23,659) compared with an event rate of 0.14% (8/5,900) in the non-contrast echo group (p=NS; one fatality per group). Other studies have reported similarly low rates of serious adverse events following echo contrast use. In a single-center study of 12,974 consecutive patients who received echo contrast, Herzog reported that the incidence of serious adverse events was 0.031% (no fatalities).25 In a multicenter study of 78,383 patients undergoing contrast echo (including more than 10,000 doses administered to critically ill or chest pain syndrome patients), Wei et al. reported the rate of serious adverse events to be 0.01% (no fatalities).26 Taken together, these data from over 100,000 patients indicate a low risk for serious adverse events (≤0.1%) in association with echo contrast administration.
The clinical safety of echo contrast has also been reported in targeted groups, including patients undergoing contrast echo within 24 hours of MI,27 as well as patients undergoing dobutamine or exercise stress testing.24,28 It is also important to recognize that rare but serious adverse events have also been reported with other imaging contrast agents, including allergic reactions to non-iodinated contrast agents used for CT29 and renal-dysfunction-associated nephrogenic systemic fibrosis with gadolinium-based contrast agents used for MRI.30
Emerging data have suggested that the benefits of contrast-enhanced imaging may extend beyond patients with poor-quality non-contrast echoes. Our group tested this concept in a study of 121 patients at high risk for thrombus due to systolic dysfunction or MI.4 Contrast echo was performed independently of the results or quality of non-contrast echo and the two echo techniques were compared with an independent standard of delayed enhancement MRI (DE-MRI) (see ‘Validation—Tissue Characterization’, below). Among this at-risk population, contrast echo yielded a near two-fold improvement in diagnostic sensitivity versus non-contrast echo (61 versus 33%; p<0.05) when both echo tests were compared with a standard of DE-MRI. Thrombi that were small in volume or mural were more likely to be missed by non-contrast echo but detected by DE-MRI (p<0.05), with a similar trend for size of thrombi missed by non-contrast but detected by contrast echo (p<0.1). When image quality scores were compared between contrast and non-contrast echo, echo contrast improved LV endocardial border definition, cavity delineation, and overall image quality versus non-contrast echo (p<0.001 for all comparisons).
However, echo performance for detection of thrombus documented by DE-MRI was not associated with reader-assigned echo quality, with similar values for all qualitative parameters between contrast and non-contrast echoes that missed thrombus versus those that detected thrombus (p=NS). Similar relationships were demonstrated when comparing image quality scores between non-contrast echoes that independently detected thrombus and those in which contrast use provided incremental utility for thrombus detection (p=NS). Patients who derived incremental diagnostic benefit from contrast administration tended to have lower LV ejection fraction (LVEF) than those in whom non-contrast echo alone accurately assessed thrombus (34 versus 41%; p=0.08), reflecting an increased prevalence of thrombus in association with impaired systolic function. Of note, in this study, among all thrombi detected by DE-MRI, 25% were non-apical in location, demonstrating the importance of comprehensive imaging for assessment of LV thrombus.
While further research is necessary to confirm the utility of an up-front strategy of contrast-enhanced imaging, these initial findings suggest that contrast-enhanced imaging should be considered for use in patients with adequate non-contrast echo image quality but high pre-test risk for thrombus.
Contrast-enhanced imaging can also be performed to assess vascular composition of cardiac masses, thereby enabling a tissue characterization based definition of thrombus to be applied. Both echo and tomographic imaging have employed this approach. Perfusion echo is generally performed during or immediately following dynamic contrast infusion, whereas DE-MRI is performed five to 30 minutes following contrast administration. Both modalities enable thrombus detection based on tissue characteristics rather than anatomical appearance, with absence of contrast uptake indicative of avascular composition. Figure 3 provides an example of improved detection of mural thrombus using tissue characterization by DE-MRI compared with anatomical imaging by non-contrast echo.
Echo, which typically applies an anatomical criterion for LV thrombus detection, has been termed the definitive clinical test for thrombus,31 with a diagnostic sensitivity believed to be approximately 90%.13,32 However, these results may be overly optimistic as most echo studies have lacked a true gold standard for thrombus. While some have included a pathology component, nearly all have been plagued by verification bias in that pathology was obtained only in selected patients. To date, we are aware of only two echo studies in which pathology verification of the presence or absence of thrombus was obtained in all patients. Visser et al.,32 studying patients with cardiomyopathy in whom a high prevalence of thrombus was anticipated, reported that non-contrast echo detected thrombus in 92% of the 26 patients with pathology-verified thrombus. However, 17% of thrombi identified by echo were false-positives and 7% of studies were excluded from analysis based on technically inadequate quality. Ezekowitz et al.,33 studying patients with LV aneurysms or degenerative mitral valve disease, reported a lower sensitivity of 77%. In this study, 25% of studies were excluded based on image quality and all patients with pathology-verified thrombus had large LV aneurysms. The relatively small number of patients in these studies, the high proportion of LV aneurysms, and the exclusion of patients based on image quality may have skewed findings concerning the ability of non-contrast echo to detect thrombus.
Several studies have used pathology data to validate tissue characterization imaging for thrombus. Kirkpatrick et al. used perfusion echo to evaluate this approach among 16 patients with cardiac masses.15 Among seven patients with thrombus (independently verified based on pathology [5/7] or resolution following anticoagulation [2/7]), all demonstrated avascular tissue properties as defined by decreased pixel intensity compared with myocardium. Similar findings were reported by Mansencal et al. in a study that employed perfusion echo to assess 31 patients with cardiac masses.16 In this study, thrombus was defined as a mass with complete lack of enhancement, consistent with avascular tissue properties. This approach enabled accurate differentiation between thrombus and neoplasm in all cases (100% accuracy versus diagnostic standard of pathology, MRI, or resolution following anticoagulation).
Validation of the tissue characterization approach has also been achieved using DE-MRI. Srichai et al. studied patients undergoing LV reconstruction surgery in whom pathology verification of presence or absence of LV thrombus was uniformly available. Cardiac MRI (including DE-MRI) was compared with echo (comprising solely anatomical imaging) among 160 patients who had both tests performed within a 30-day interval.2 Sensitivity of MRI was 88%, compared with 23% for transthoracic echo. These results are in general agreement with our group’s findings in a study that compared DE-MRI with cine-MRI.3 Among 784 consecutive patients with systolic dysfunction, the prevalence of LV thrombus was higher by DE-MRI versus cine-MRI (7 versus 4.7%; p<0.005). Thrombus detected by DE-MRI but missed by cine-MRI was typically mural in shape or small in size (p<0.05). Pathology findings and follow-up data were consistent with DE-MRI as a better reference standard for thrombus versus cine-MRI. Patient stratification according to presence or absence of thrombus by DE-MRI yielded over a seven-fold difference in validative end-points (transient ischemic attacks [TIA], cerebrovascular accident [CVA], or pathology-verified thrombus) between groups (15.1 versus 2.1%; p<0.0001), including accurate assessment for thrombus in all patients (eight/eight) with pathology verification.
Further stratification based on cine-MRI did not improve differentiation of patients concerning thrombo-embolic events or pathology verification of thrombus. Taken together, these DE-MRI and perfusion echo studies demonstrate that contrastenhanced tissue characterization provides a highly accurate approach to LV thrombus assessment.
Predisposing Risk Factors
Improved assessment of thrombus by contrast-enhanced imaging has yielded new insights regarding clinical and imaging variables that predispose to thrombus formation.
Contractile Dysfunction and Remodeling
LV contractile dysfunction and aneurysmal remodeling are well-established markers for thrombus formation. Multiple echo studies, typically using non-contrast imaging, have reported that risk for thrombus has been shown to be proportional to severity of contractile dysfunction. Global indices of contractile dysfunction such as ejection fraction9,11 and wall motion score10,34–36 have been linked to thrombus. LV aneurysms37 and dyskinetic regions5,38 have also been reported to be more common in patients with thrombus. These associations likely reflect a link between contractile dysfunction and blood stasis, with stasis producing a thrombogenic mileu within the LV cavity.
Clinical studies have also reported that prevalence of thrombus is increased in patients with large myocardial infarctions. Enzymatic infarct size9,36,38 and absence of vessel patency39 have been associated with increased prevalence of thrombus. LV thrombus has also been reported to be more prevalent following anterior MI,11,12,39 suggesting that infarct distribution modifies risk for thrombus. However, all of these studies used indirect markers of infarct size and location, such as enzyme levels or angiographic indices, and employed a separate test (typically echo) to assess contractile function. Thus, while an association between infarct size and thrombus was demonstrated, the mechanism had commonly been assumed to be more extensive systolic dysfunction and/or remodeling, and not necessarily the presence of infarction.
MRI provides highly accurate assessment of infarct size, contractility, and thrombus within a single imaging examination. Cine-MRI enables highly reproducible assessment of LV contractile function with excellent endocardial cavity definition.40,41 DE-MRI provides non-invasive infarct imaging with near exact replication of pathology-evidenced infarct size and morphology.42,43 Cine- and DE-MRI are typically performed during the same imaging session. This has facilitated study of the relative roles of LV geometry, contractile function, and infarct size as structural predictors of thrombus.
Recent MRI studies have demonstrated that infarct size is linked to risk for thrombus. In a study of 57 patients with acute MI or established CAD, Mollet et al. reported that patients with DE-MRI-evidenced thrombus had larger infarct size, lower ejection fraction, and more frequent LV aneurysms.1 However, the independent predictive value of infarct size (after controlling for contractile dysfunction or aneurysmal dilation) was not assessed in this study. Data from our group have demonstrated that risk for thrombus is independently linked to infarct size, contractile dysfunction, and aneurysmal remodeling, with each parameter providing additive predictive value. In our initial comparative study of DE- and cine- MRI among heart failure patients,3 LV thrombus was independently associated with infarct size even after controlling for age, CAD, and ejection fraction. Thrombus prevalence was five-fold greater in patients with ischemic compared with non-ischemic cardiomyopathy (9.2 versus 1.7%; p=0.0002) despite similar mean ejection fraction (31.8 versus 31.7%; p=0.88) in the two groups, with increased prevalence of thrombus paralleled by increased prevalence of infarcted myocardium in patients with ischemic cardiomyopathy. The association between infarct size and thrombus has also been demonstrated in research that has employed echo to measure LV geometry and contractility. In our subsequent study of echo and MRI,4 thrombus was again associated with infarct size by DE-MRI even after controlling for LVEF and aneurysmal dilation as measured by either echo or cine-MRI. These results demonstrate that infarct size is an independent structural marker for thrombus irrespective of the modality used to assess contractile function.
The association between thrombus and infarct size is supported by pathology studies. Using a canine infarct model with post mortem histopathological verification of scar, Jugdutt et al. reported that thrombus prevalence was increased among dogs with Q-wave compared with those with non-Q-wave MI (64 versus 0%; p<0.005).44 Infarct size was larger and LV remodeling more advanced in the Q-wave group, paralleling increased prevalence of thrombus. Clinical studies in the post-MI setting have found thrombus to be associated with larger infarcts.9,39 While the mechanism relating scar to thrombus is unknown, this may be secondary to pro-thrombotic endocardial alterations or subtle differences in LV contraction between regions of infarction and those of dysfunctional but viable myocardium.
Consensus guidelines recommend that contrast-enhanced echo be used to confirm or exclude the diagnosis of thrombus when non-enhanced images are sub-optimal for definitive diagnosis.21 However, as detailed above, emerging data suggest that contrast-enhanced imaging may have broader utility as thrombus can be missed by non-contrast echo even when image quality is judged to be optimal.4 One alternative approach to thrombus screening concerns stratification based on risk factors such as infarct size or contractile function, with an up-front strategy of contrast-enhanced tissue characterization imaging used for patients at high pre-test risk for thrombus. In research settings, such a strategy has yielded over a two-fold increase in thrombus detection versus non-contrast echo. Future studies are necessary to assess the clinical implications and costs associated with a primary strategy of contrast-enhanced thrombus imaging based on clinical profiles rather than echo image quality.
Contrast-enhanced echo markedly improves anatomical detection of LV thrombus versus non-contrast echo and the diagnostic benefits of contrast use have been associated with improved resource utilization, altered therapeutic management, and overall cost savings.20 Contrast can also be used for tissue characterization, which provides further benefit for thrombus detection versus anatomical imaging.2,4,15,16
Both perfusion echo and DE-MRI have been used to detect thrombus based on avascular tissue characteristics; this approach has been shown to be superior to anatomical imaging for each of these modalities. Improved thrombus detection has facilitated identification of structural risk factors for thrombus formation: infarct size, contractile dysfunction, and aneurysmal remodeling.