Spontaneous echocardiographic contrast (SEC) or “smoke” in the left atrium (LA) is a pattern of increased blood echogenicity caused by ultrasonic backscatter from red blood cell aggregates. These aggregates form due to noncovalent binding between red cells and plasma proteins under low flow and low shear conditions (1- 3). Since this interaction is largely independent of platelet and coagulation cascade, it is unaltered by the administration of antiplatelet and anticoagulant drugs (2- 3). Even though stasis of blood in the LA is the most significant factor contributing to SEC, clinical studies have demonstrated correlation with hemorheological abnormalities such as increased hematocrit and increased fibrinogen levels (3- 4).

Several studies have shown that SEC is a frequent finding in patients with atrial fibrillation (AF), mitral stenosis and in patients with prior thromboembolism, occurring in about 30% to 60% of such patients, depending on clinical circumstances (5- 15). The clinical importance of SEC in such patients is its association with LA thrombus and increased thromboembolic complications. Most of these studies, however, involved patients with AF; there is very little data on the significance of SEC in sinus rhythm (NSR). The purpose of our study was to define the clinical and echocardiographic characteristics of patients with left atrial spontaneous echo contrast (LASEC) in NSR and to evaluate its clinical significance in relation to LA thrombus and cardiac embolism.

We analyzed the reports of 1,288 consecutive transesophageal echocardiograms (TEE) performed between the years 1993 through 1997. All TEE were performed using Omniplane 5 MHz probes on Hewlett Packard Sonos 1000 echocardiographs, according to recognized guidelines (16). Echocardiograms with SEC were selected and reviewed. Patients with swirling SEC in the LA who were in NSR during TEE were chosen for analysis. Patients in AF or flutter at the time of TEE were excluded from this study. Left atrial SEC was diagnosed by the presence of swirling dynamic smoke-like echoes within the atrial cavity that was distinct from white noise artifact. The presence and severity of SEC were determined independently by two different echocardiographers, and any discrepancy was resolved by consensus. Spontaneous echo contrast was graded as 0 = no smoke, 1+ = mild smoke visible in some portion of the LA, 2+ = dense smoke that appeared throughout the LA. The presence of LA thrombus, diagnosed as a clearly defined echo-dense intracavitary mass not due to the pectinate muscles of the LA appendage (LAA), was specifically sought. Peak LAA emptying velocity (LAAEV) was measured using pulsed Doppler with the sample volume placed 1 cm inside the orifice of LAA. Visualization of the thoracic aorta was routinely performed, and any protruding atheromas ≥4 mm were noted.

A two-dimensional and Doppler transthoracic echocardiogram performed before TEE was analyzed in all patients. Standard acoustic windows had been obtained using a 2.5 MHz transducer. On the apical four-chamber view, the LA size was measured in two dimensions, length (anteroposterior dimension) and width (lateral dimension). Ventricular function and ejection fraction were visually evaluated and qualitatively assessed as normal, mild, moderate or severe dysfunction. The presence or absence of left ventricular (LV) hypertrophy was determined, and mitral inflow velocities E, A and E:A ratio were measured. Since we hypothesized that diastolic dysfunction might predispose to LA stasis, mitral inflow velocities were used to evaluate LV diastolic dysfunction and were categorized as impaired relaxation, pseudonormalization or restrictive patterns (17). Standard criteria were used to quantify mitral regurgitation as mild, moderate or severe (18).

Clinical characteristics of study patients were obtained from patient records. The clinical data collected included age, gender and traditional cardiovascular risk factors—hypertension, diabetes mellitus, smoking and hypercholesterolemia. We noted the presence of clinical coronary artery disease, congestive heart failure (CHF), significant valve abnormalities including prosthetic valves, past history of paroxysmal AF or flutter, stroke or transient ischemic attack, peripheral emboli and carotid stenosis. In all patients the clinical indication necessitating the performance of TEE was also noted.

In order to better define the significance of smoke in NSR and its relation to thromboembolism, a control group of 45 patients with no SEC were randomly selected from the initial 600 consecutive TEE studies. Controls were age matched to the case group, had no SEC, were in NSR, had LA dilation >4.0 cm on M-mode and no significant mitral regurgitation. These characteristics were chosen to attempt to match baseline characteristics of the two groups, except for the presence of SEC. We chose patients with no significant mitral regurgitation (MR) because the presence of significant MR reduces the occurrence of SEC in the LA (19). We chose patients with large LA for the control group to attempt to exclude the influence of LA size, to isolate the effect of SEC. In a fashion similar to the case group, clinical information, transthoracic and TEE variables were obtained.

After obtaining data from both the case and control groups, the results were sequentially analyzed in the following steps. First, we defined the characteristics of patients with SEC in NSR. Next, we compared the case group with a control group of patients with no SEC. Since patients with SEC had other factors that can be associated with cerebrovascular accident (CVA), such as increased LA size (20) and decreased LAAEV, we decided to further analyze our data by combining both the case and control groups together. Then we compared patients with CVA and no CVA to determine the factors that were independently associated with CVA.

Statistical analyses were performed using SPSS version 6.1 (SPSS Inc., Chicago, Illinois). Variables are presented as mean ± SEM. Student t test was used to test the difference in the means of variables between patients with and without SEC who had TEE. Mantel-Haenszel chi-square was calculated to test whether the indicator variables were independent of each other. A forward stepwise method was used to select four significant variables with p ≤ 0.05 for entry into a multivariate regression analysis. Similar analyses were then performed to evaluate the differences between patients with and without CVA. Then, a multivariate logistic regression analysis (with stepwise method) was performed on this combined group using all echocardiographic variables with p ≤ 0.25. The relationship between SEC and CVA was also tested by a logistic regression analysis with LA size as a covariate. In all analyses, a two-tailed p value of ≤0.05 was considered significant.

Of 1,288 consecutive TEE studies performed, 195 patients (15%) had swirling SEC in the LA. Of these, 24 patients (12%) were in sinus rhythm and formed our study group. This represented 2% of our overall TEE population. Among the 1,288 TEE studies, 50% were referred for TEE to detect a source of CVA.

The clinical and echocardiographic characteristics of the patients with LASEC are depicted in (Tables le1, le2), respectively. The mean age was 72 years. There was a high prevalence of manifest cardiac disease. Coronary artery disease was present in 58%, CHF in 33%, valve stenosis in 12% (two mitral stenosis and one aortic stenosis), prosthetic valves in 21% (two prosthetic aortic valves, one prosthetic mitral valve and two patients with both) and a past history of AF or flutter in 25%. Hypertension was present in 71%. Past history of CVA was noted in 83%.

Spontaneous echo contrast was 2+ in 83% of the study group. All patients with SEC had significantly dilated LA. The mean LA size was 5.6 ± 0.6 cm. The mean LAAEV was reduced, 38 cm/s (median 35); 58% had a decreased emptying velocity, ≤40 cm/s. Left atrial thrombus was noted in 13%. Sixty-seven percent of patients had normal LV systolic function, while 25% had severe and 8% had mild to moderate LV dysfunction. Left ventricular thrombus was noted in 4%. Forty-one percent of patients had LV hypertrophy. Mitral inflow velocities consistent with diastolic dysfunction were noted in 33% of patients; 12% had a restrictive pattern and 21% had impaired relaxation or pseudonormalization. No patient with SEC had moderate or severe MR. Severe atherosclerosis of the aorta with protruding atheromas ≥4 mm was noted in 38% of patients. Conditions that could predispose to stasis of blood within the LA such as reduced LAAEV, mitral stenosis, systolic or diastolic LV dysfunction and a past history of AF or flutter were present in 92% of patients. However, two patients (8%) had LA dilatation as their sole abnormality with none of the above mentioned factors.

Our control group was selected to have no SEC, NSR, dilated LA >4.0 cm and no significant MR. We selected such a control population to try to isolate SEC as a single variable. Control patients also had a high prevalence of manifest cardiac disease. Comparison of the two groups (Tables le1, le2) revealed no significant differences in the prevalence of cardiovascular risk factors, coronary artery disease, CHF, prosthetic valves, history of AF or LV dysfunction. A high prevalence of CVA was noted in both the groups. This reflects the referral bias intrinsic to TEE studies. Despite this referral bias, a higher prevalence of CVA was noted in patients with SEC as compared with the control group, 83% versus 56%, p = 0.02. Patients with SEC had a larger LA, 5.6 versus 4.9 cm, p < 0.0001. Patients with SEC had a lower mean LAAEV, 38 versus 56 cm/s, p = 0.001. Fifty-eight percent of patients with SEC had LAAEV <40 cm/s compared with only 5% of controls. Importantly, 13% of patients with SEC had LA thrombus as compared with none in the control group, p = 0.02.

By multivariate analysis, CVA, larger LA size and decreased mean LAAEV were significantly associated with SEC in NSR (Table le3). Even after adjusting for LA size, patients with SEC had higher prevalence of stroke than controls, p = 0.03.

Patients with SEC have other factors that can be associated with CVA, such as increased LA size and decreased LAAEV. Therefore, the data was further analyzed by combining case and control groups together to compare patients with CVA and no CVA to determine which factors correlate independently with stroke. (Tables le4, le5) present these data.

Spontaneous echo contrast in LA was more prevalent in patients with stroke as compared with patients with no stroke, 44% versus 17%, p = 0.02. Among all the clinical and echocardiographic variables, only SEC was significantly associated with CVA. Patient age and other risk factors were similar between both groups. History of AF and LV function was comparable. Though more patients with CVA had protruding atheromas, 40% versus 25%, this difference did not reach statistical significance. The mean LA size (5.0 vs. 5.1 cm) as well as the mean LAAEV (48 vs. 54 cm/s) was not significantly different between the two groups. By the multivariate analysis, including SEC, LAAEV, ejection fraction, LA thrombus and protruding aortic atheromas, SEC was the only independent correlate of CVA, p = 0.026.

Most prior studies of LASEC have been of patients with AF or mitral stenosis. The prevalence of SEC varies depending on the patient population studied. In AF, SEC occurs in 45% to 60% of patients. In AF patients, SEC is related to large LA dimension and decreased atrial mechanical function and is associated with increased risk of LA thrombus and embolic complications. Stroke or embolic event rate is 12% per year in AF patients with SEC, compared with 3% per year in AF patients with no SEC. Left atrial thrombus occurs in 12% to 24% of AF patients with SEC (5,13- 14).

In patients with mitral stenosis regardless of rhythm, SEC is noted in 65% to 80% of cases. Again, the presence of SEC is particularly associated with large LA size and increased risk of thromboembolism. Left atrial thrombus is detected in 18% to 23% of patients with mitral stenosis (6- 7,9,15). In patients with acute CVA or chronic cerebrovascular disease, SEC occurs in 32% to 38% and is associated with increased risk of LA thrombus (8,10). These data effectively establish the prothrombotic nature of SEC and its strong association with clinical thromboembolism.

In this study, LASEC was found in 15% of all patients undergoing TEE. In the group with SEC, 12% had smoke during NSR. Thus, the prevalence of SEC in NSR among our patients who were referred for TEE is about 2%. These findings are very similar to the results of Black, who described an SEC prevalence of 19% among 400 consecutive patients undergoing TEE. Similar to our findings, 13% of patients with SEC in their study were in NSR, 2.5% of their entire group (13). However, in that study, 6 of 10 patients with SEC in NSR were found to have mitral stenosis, while our larger study had a much lower prevalence of mitral stenosis (8%). A previous study of SEC in NSR included only patients with mitral stenosis; SEC was noted in 45% of patients (9).

Our study for the first time describes the clinical and echocardiographic correlates of LASEC in NSR in a large patient population referred for TEE. Patients with LASEC had a large prevalence of clinically manifest heart disease—coronary artery disease, CHF, paroxysmal AF and valve stenosis. But, when we compared the SEC group to age matched controls, we found that the only independent predictors of SEC were prior CVA, large LA size and decreased mean LAAEV (Tables le1, le2). Left atrial size in patients with SEC was even larger than the control group preselected for LA size >4.0 cm. Though we had hypothesized a priori that history of paroxysmal AF, presence of severe LV dysfunction, LV hypertrophy and LV diastolic dysfunction might correlate with SEC, these variables were no different between the case and control groups.

We did find a decrease in mean peak LAAEV in the LASEC group indicating impaired atrial appendage mechanical function. More patients with SEC in NSR had decreased LAAEV, suggesting that a depressed atrial contractile function is a causative factor in the development of SEC in NSR. In light of the high prevalence of manifest cardiac disease, fixed atrial contractile dysfunction may be the largest contributor to decreased LAAEV; however, we cannot exclude that the SEC group had more undetected AF and LA stunning (21). Consistent with the prior observations that turbulent flow originating from significant MR may protect against the formation of SEC, no patients in the LASEC group had significant MR (19).

An important observation in our study was that 13% of patients with SEC had LA thrombus despite NSR. No patient in the control group had LA thrombus. This prevalence of thrombus is similar to the 12% to 24% prevalence noted in patients with AF and SEC (4- 5,14). Further, a higher prevalence of CVA was noted in our study patients compared with control patients without SEC. Moreover, when cases with SEC were combined together with controls, SEC was the only independent predictor of CVA. These observations indicate that SEC represents a prothrombotic state even in patients with NSR and that the high prevalence of LA thrombus may be responsible for the higher stroke rates.

Our results indicate that LASEC is a more powerful correlate of stroke than depressed LAAEV. Comparison of patients with CVA and no CVA revealed no difference in mean LAAEV (48 cm/s vs. 54 cm/s); only prevalence of SEC was significantly different (44% vs. 17%) p = 0.02. Recently, attention has focused on the association between depressed LAAEV, LA thrombus and stroke. Tabata et al. (22) showed that patients in NSR with high pulmonary venous wedge pressure had significantly decreased LAAEV and larger LAA area. Among the 11 patients in NSR with high wedge pressure, one had LA thrombus and two had SEC. These observations suggest that, even in patients with NSR, a marked elevation of LA pressure can reduce LAAEV and predispose to thrombus formation. Kamalesh et al. (23) reported two patients with embolic stroke in NSR and depressed LAAEV. Importantly, both patients had SEC and LA thrombus. LAA function in stroke patients with NSR was depressed compared with control patients with NSR with no stroke and was similar to the depressed LAA function in patients with chronic AF. Thus, the triad of SEC, depressed LAAEV and LA thrombus has been previously reported in patients with NSR and stroke. Our data suggest that though depressed LAAEV is a cause for SEC, it is the latter that correlates more highly with CVA.

Several clinical and epidemiological studies have shown an association between increased fibrinogen levels and stroke and cardiovascular mortality (24- 26). Besides other mechanisms, increased fibrinogen levels predispose to the formation of LASEC; positively charged fibrinogen acts as a bridge between negatively charged red cells, facilitating aggregation (1- 3). An underlying condition that may lead to stasis of blood in the LA was noted in 92% of our patients with SEC. However, two patients with SEC (8%) in NSR had dilated LA as the sole abnormality. In AF patients with SEC, higher fibrinogen levels are found (4). We advance the untested hypothesis that higher fibrinogen levels may be found in SEC patients in NSR as well. This might explain SEC in the absence of any abnormality but LA dilation. These two patients did not have depressed LAAEV, so undetected AF is not a likely cause.

Study design: our study was an observational retrospective study of a selected group of patients from a referred TEE population. However, our results on the overall prevalence of LASEC is similar to that obtained in several other studies involving a wide variety of patients. Further, the prevalence of SEC in NSR is similar to the prevalence obtained by analysis of a previous smaller study (13). Control patients, though they have LA dilation, were not matched for the magnitude of LA dilation found in the SEC group. Though control group patients had LA dilation >4.0 cm, there were few patients, even in this large clinical experience, with the high degree of LA dilation found in the SEC group, who were also in NSR, with no significant MR and no SEC. Most patients with this large LA size and no SEC were in AF or had significant MR. Even after adjusting for LA size by covariate analysis, patients with SEC had a higher prevalence of stroke than controls. Since SEC in NSR is uncommon, the number of study patients was relatively small, and, hence, the results should be interpreted with caution. The small sample size may explain the lack of association between CVA and aortic plaques. Prospective follow-up of patients was not performed.

Limitations pertaining to the cause of LASEC: ambulatory electrocardiogram monitoring of patients with LASEC in NSR was not done but would be of interest to ascertain whether undetected paroxysmal AF is a cause. Systematic analysis of diastolic function of the LV was not performed; diastolic dysfunction was assessed with mitral inflow velocities and not pulmonary vein velocities in all patients. Hence, we cannot exclude that our patients had a higher prevalence of diastolic dysfunction than the 33% noted in our study population. We did not measure serum levels of fibrinogen or other plasma proteins to correlate with SEC in NSR. Future studies addressing this limitation would be of interest.

The data in this study indicate that LASEC in NSR is a distinct clinical phenomenon occurring in 2% of the referred TEE population. It occurs in patients with significantly dilated LA. Its association with decreased LAAEV indicates that depressed atrial contractile function is also a causative factor.

Left atrial thrombus occurs in a significant proportion of patients with LASEC despite NSR. We hypothesize that a combination of increased LA size and decreased LA mechanical function from whatever cause predispose to SEC and thrombus formation. We have found an association between SEC in NSR and a higher prevalence of CVA. Further, SEC is found to be an independent and more powerful correlate of CVA than reduced LAAEV. The data in this study indicate that LASEC in NSR is a prothrombotic condition. A natural history study to prospectively follow such patients would be of interest.

A higher prevalence of LA thrombus and stroke even in NSR raises another question: whether patients with SEC in NSR would benefit from anticoagulation. Spontaneous echo contrast in a given patient provides additional objective evidence of significantly depressed atrial mechanical function, LA stasis and a prothrombotic milieu that might warrant anticoagulation therapy. A possible treatment benefit from anticoagulation in such patients warrants study as well.

The authors would like to thank Mrs. Carmen Stockfisch for her work in the preparation of this manuscript.