Predictors of Systemic Embolism in Patients with Mitral Stenosis: A Prospective Study

Methods

Patients

Eligible patients were consecutive adults (age ≥ 15 years) with mitral stenosis (mitral valve area ≤ 2 cm² according to echocardiographic planimetry) who presented to a university-affiliated medical institution from April 1987 to December 1994. We excluded patients with infective endocarditis and those who were in critical condition because of systemic embolism and died during hospitalization.

End Point

The study end point was the occurrence of new systemic embolism during follow-up. The diagnosis of systemic embolism was based on symptoms and signs (sudden onset of peripheral arterial ischemic [for example, sudden flank pain with hematuria, abdominal pain with gastrointestinal bleeding, or leg pain with pulse deficit] or neurologic manifestations without prodromes) and on findings from computed tomography, angiography, and surgery. We did not attempt to detect silent emboli.

Clinical Variables

We assessed nine clinical variables (Table 1): age at enrollment; sex; presence or absence of previous systemic embolism, atrial fibrillation, hypertension, and New York Heart Association class III or IV congestive heart failure; and therapy with anticoagulants, percutaneous balloon mitral commissurotomy, or valvular surgery. Patients were regularly followed at outpatient clinics.

Echocardiographic Method and Variables

Standard transthoracic echocardiography was done at enrollment in all patients by using a Hewlett-Packard 7340, Sonos 1000, or Sonos 1500 echocardiographic system (Hewlett-Packard, Palo Alto, California) interfaced with both 2.5-MHz and 5.0-MHz transducers. Biplane or omniplane transesophageal echocardiography using a 5.0-MHz transducer was also performed in a subgroup of consecutive patients who entered the study from September 1991 to October 1992. Ten echocardiographic variables were examined (Table 1). Mitral valve area was measured by planimetry from two-dimensional echocardiography. When two-dimensional echocardiography of the mitral orifice yielded unsatisfactory results, we used the pressure half-time (T_1/2) method (mitral valve area [cm²] = 220/T_1/2 ms) [10-12]. We did not use the pressure half-time method when the mitral orifice could be clearly defined by two-dimensional echocardiography because pressure half-time is influenced by many factors other than mitral valve area [13, 14]. Other echocardiographic variables were left atrial diameter at end systole; presence or absence of a left atrial thrombus [15] or left atrial smoky echoes on transthoracic or transesophageal echocardiography [16-18]; presence or absence of impaired left ventricular systolic performance; and presence or absence of significant (moderate or severe) aortic stenosis, aortic regurgitation, mitral regurgitation, tricuspid regurgitation, or pulmonic regurgitation.

The degrees of these valvular lesions were semiquantified by using a continuity equation (for aortic stenosis) or color Doppler imaging (for various regurgitations), as described elsewhere [19, 20]. Briefly, significant aortic stenosis refers to an aortic valve area of 1.2 cm² or less determined by the continuity Equation method RF 19*; significant mitral or tricuspid regurgitation refers to a ratio of regurgitant jet area to left or right atrial area of 20% or more; and significant aortic or pulmonic regurgitation refers to a ratio of jet width to ventricular outflow tract diameter of 25% or more [20]. Left atrial diameter at end systole was measured from an M-mode echocardiogram recorded in parasternal long-axis view. The measurement was made according to the recommendations of the American Society of Echocardiography [21]. To detect left atrial smoky echoes, we used a 5-MHz transducer during transthoracic and transesophageal echocardiography because a 5-MHz transducer is more sensitive than a 2.5-MHz transducer [16].

We chose the term smoky echoes instead of spontaneous echocardiographic contrast (a term frequently used in other studies [6, 7]) because some patients with severe tricuspid regurgitation or right heart failure had bright moving spots (originating from microbubbles) in the venae cavae or hepatic veins that were identical to those seen during contrast echocardiography. Thus, we reserve spontaneous echocardiographic contrast for that echocardiographic pattern and use smoky echoes for the finer, lighter whorling echoes (originating from aggregated erythrocytes) [16-18] that appeared in the left atrium in patients with severe mitral stenosis.

Statistical Analysis

For each clinical and echocardiographic measure, the log-rank statistic was used to determine whether the overall pattern of the time to development of systemic embolism (embolism-free time) varied among levels of the measure. Mean embolism-free time was estimated by using a nonparametric method that considers censoring [22]. Cox regression was used to examine the significance of the clinical and echocardiographic variables in predicting embolism-free time for patients in sinus rhythm and patients in atrial fibrillation. All analyses were performed by using BMDP Dynamic Release 7.0 [23].

Results

Five hundred thirty-four patients were followed for a mean (±SD) of 36.9 ± 22.5 months. Of these, 257 patients (48.1%) received anticoagulants throughout the follow-up period. The indications for anticoagulation were the presence of a left atrial thrombus, atrial fibrillation, or a history of systemic embolism; patient compliance with therapy; and lack of risk factors for bleeding. The relatively low percentage of patients receiving anticoagulants in this series was due to minimal patient compliance. During the follow-up period, 60 patients (11.2%) developed a systemic embolism.

When Cox regression was performed, significant interaction was found between atrial fibrillation and age, percutaneous balloon mitral commissurotomy, mitral valve area, previous systemic embolism, left atrial thrombus, and anticoagulation. In other words, the significance of these variables depended to some extent on whether the patient was in atrial fibrillation or sinus rhythm. We therefore performed subgroup analyses.

Subgroup Analyses

Of the 132 patients in sinus rhythm, 12 (9.1%) developed systemic embolism during follow-up. Age (P < 0.001), percutaneous balloon mitral commissurotomy (P = 0.02), and mitral valve area (P = 0.02) were significant predictors in the log-rank analysis (Table 2). Results of the Cox regression showed that age (relative risk [RR], 1.12 [95% CI, 1.04 to 1.21]), left atrial thrombus (RR, 37.1 [CI, 2.82 to 487.8]), mitral valve area (RR, 16.9 [CI, 1.53 to 187.0]), and significant aortic regurgitation (RR, 22.4 [CI, 2.72 to 184.8]) were significant predictors of new systemic embolism (Table 3). No interactions were found among these variables. However, mitral valve area became a nonsignificant predictor (P = 0.12) when patients with percutaneous balloon mitral commissurotomy were excluded from the analysis.

Of the 402 patients in atrial fibrillation, 48 (11.9%) developed systemic embolism. Age (P = 0.01), previous embolism (P = 0.001), and percutaneous balloon mitral commissurotomy (P = 0.003) were significant predictors in the univariate analysis (Table 2). In the multivariate analysis, however, only previous embolism (RR, 3.11 [CI, 1.66 to 5.85]) and percutaneous balloon mitral commissurotomy (RR, 0.37 [CI, 0.18 to 0.79]) remained significant predictors of embolism-free time (Table 3). Again, no interactions were found between these two variables.

A subgroup analysis of the 164 patients who underwent baseline transesophageal echocardiography revealed no other significant predictors.

Discussion

Factors Correlated with Systemic Embolization in Patients with Mitral Stenosis

Our prospective study revealed that for patients in sinus rhythm, embolization was related to age, mitral valve area, the presence of a left atrial thrombus, and significant aortic regurgitation. For patients in atrial fibrillation, the significant factors were previous embolism and percutaneous balloon mitral commissurotomy (Table 3). Other retrospective studies have shown that atrial fibrillation [1, 2, 8], age [1, 2, 8], and previous embolism [3] correlate with increased incidence of systemic embolism in patients with mitral stenosis and that age is closely related to the prevalence of atrial fibrillation [24] and to a history of embolization [1, 2, 8]. Several studies have shown that anticoagulation reduces the incidence of systemic embolism in patients with mitral stenosis and atrial fibrillation [25-27]. To the best of our knowledge, however, our study is the first to show that the presence of a left atrial thrombus and significant aortic regurgitation are positive predictors and that percutaneous balloon mitral commissurotomy seems to be a negative predictor.

Left Atrial Thrombus and Systemic Embolism in Patients with Mitral Stenosis Who Are in Sinus Rhythm

Dislodgement of a left atrial thrombus in patients with mitral stenosis has been thought to lead to systemic embolism. Although a correlation between left atrial thrombus and systemic thrombus would be expected, previous studies have not confirmed such a correlation. We found that the presence of a left atrial thrombus was a positive predictor (RR, 37.1 [CI, 2.82 to 487.8]) for patients in sinus rhythm (Table 3). It is generally accepted that anticoagulants should be given to patients with mitral stenosis who are in atrial fibrillation [25-27], but the appropriate use of these agents in patients with mitral stenosis who are in sinus rhythm has been uncertain. Our findings suggest that anticoagulation would be appropriate at least for patients with a left atrial thrombus shown by two-dimensional echocardiography.

Aortic Regurgitation and Systemic Embolism in Patients with Mitral Stenosis Who Are in Sinus Rhythm

We found that the presence of significant (moderate or severe) aortic regurgitation was a positive predictor (RR, 22.4 [CI, 2.72 to 184.8]) of systemic embolism in patients with mitral stenosis who are in sinus rhythm. One possible explanation is that because significant aortic regurgitation tends to increase left ventricular diastolic pressure, the diastolic pressure gradient between the left atrium and the left ventricle would be decreased. This sets the stages for reduction of transmitral blood flow, aggravation of left atrial blood stasis, predisposition to left atrial thrombus formation, and systemic embolism.

Mitral Valve Area and Systemic Embolism in Patients with Mitral Stenosis Who Are in Sinus Rhythm

One would expect that the smaller the mitral valve area, the higher the incidence of systemic embolism. A first glance at Table 3 may lead one to think that our study showed the contrary-that patients with a larger, rather than smaller, mitral valve area seemed to have a high risk for systemic embolism. However, a more careful examination of the data shows that the significance of mitral valve area in patients in sinus rhythm could be a false alarm. Patients with a small mitral valve area were more likely to have had percutaneous balloon mitral commissurotomy. When patients who underwent percutaneous balloon mitral commissurotomy were excluded, mitral valve area became a nonsignificant predictor (P = 0.12). Another implication of this finding could be that the prevention of systemic embolism in patients with mitral stenosis may need to be forestalled at the earliest opportunity as long as the mitral valve area is 2 cm² or less. Further studies are needed to clarify this issue.

Percutaneous Balloon Mitral Commissurotomy and Systemic Embolism in Patients with Mitral Stenosis Who Are in Atrial Fibrillation

Percutaneous balloon mitral commissurotomy has become one of the standard, and often first-line, treatments for patients with mitral stenosis [28-32]. It can substantially alleviate symptoms [31]. In our study, we were encouraged to find that percutaneous balloon mitral commissurotomy was a negative predictor of systemic embolism in patients with mitral stenosis who were in atrial fibrillation (Table 3). We speculate that in some patients, the presence of an interarterial shunt might lead to pulmonary embolism instead of systemic embolism when a left atrial thrombus becomes dislodged. Amelioration of regional left atrial hypercoagulability by balloon mitral commissurotomy may also play a role [33]. The identification of percutaneous balloon mitral commissurotomy as an independent negative predictor of systemic embolism favors the early use of this procedure in patients with mitral stenosis who are in atrial fibrillation. Because the procedure seemed to be an independent negative predictor, it would help patients regardless of their anticoagulation status. Furthermore, earlier intervention may also be beneficial in patients in sinus rhythm to prevent the development of atrial fibrillation and systemic embolism.

Left Atrial Smoky Echoes and Systemic Embolism in Patients with Mitral Stenosis

Previous retrospective studies have reported that the presence of left atrial smoky echoes by transesophageal echocardiography might favor the occurrence of systemic embolism in patients with mitral stenosis [6, 7]. Our prospective study of a subgroup of 164 patients who underwent baseline transesophageal echocardiography did not confirm this finding. Nevertheless, we found that the significance of previous embolism was just above the 0.05 limit (P = 0.056) when left atrial smoky echoes were present. This implies that left atrial smoky echoes might confound or interact with other variables. However, only 11 patients in this subgroup had embolism; larger studies are needed to clarify the issue.

Clinical Implications

We discovered important predictors of systemic embolism in patients with mitral stenosis; these findings suggest that anticoagulation may be prudent not only in patients in atrial fibrillation and patients with a previous systemic embolism but also in patients showing significant aortic regurgitation or left atrial thrombi on echocardiography. Therefore, echocardiography should be carefully performed to look for any evidence of left atrial thrombus or significant aortic regurgitation in patients who are in sinus rhythm and have no history of systemic embolism. Our study also suggests that early percutaneous balloon mitral commissurotomy may be another effective measure (regardless of anticoagulation status) to help prevent systemic embolism in patients with mitral stenosis.

From Chang Gung Memorial Hospital and Chang Gung University, Taipei, Taiwan, Republic of China; and Deaking University, Burwood, Victoria, Australia.

Dr. Lo: School of Public Health Faculty of Health and Behavioral Science, Deakin University, 221 Burwood Highway, Burwood, Victoria 3125, Australia.