Stroke and systemic embolism occur frequently in patients with rheumatic mitral stenosis (MS) in sinus rhythm (SR), but the risk and predictors of embolic events in this population are not well studied. The aim of this study was to determine if transient, subclinical atrial fibrillation (AF) increases the risk of systemic embolism in patients with MS in SR. A single-center, prospective observational study of patients with rheumatic MS in SR was performed. The rate of the composite primary outcome of stroke, transient ischemic attack, or non–central nervous system embolism was determined, as well as the predictive value of Holter-detected episodes of transient (<30 seconds), subclinical AF for this outcome. Hazard ratios were derived for subclinical AF, after adjustment for clinical and echocardiographic predictors of systemic embolism, using Cox regression. The sensitivity, specificity, and area under the receiver-operating characteristic curve of subclinical AF were determined for the primary outcome. Among 179 patients (mean follow-up 10.2 months), the rate of the primary outcome was 5.3/100 patient-years (95% confidence interval [CI] 2.6 to 10.5). In univariate analysis, subclinical AF (hazard ratio 4.54, 95% CI 1.08 to 19.0, p = 0.038) and dense spontaneous echocardiographic contrast (hazard ratio 4.32, 95% CI 1.03 to 18.09, p = 0.045) were predictors of the primary outcome. In multivariate analysis, subclinical AF remained the only significant predictor (hazard ratio 5.02, 95% CI 1.15 to 22.0, p = 0.032). Subclinical AF had an area under the receiver-operating characteristic curve of 0.68 and high negative predictive value (97.7%) for the primary outcome. In conclusion, Holter-detected, transient (<30 seconds), subclinical AF is a predictor of stroke and systemic embolism in patients with rheumatic MS in SR. Considering the high risk for embolism, randomized trials of oral anticoagulation are needed in this population.
Highlights
- •
Rheumatic MS is a common cause of stroke and systemic embolism in developing countries.
- •
Although most patients who experience strokes are in AF, about 1/5 are in sinus rhythm. There is little information on the risk for and predictors of stroke and systemic embolism in these patients.
- •
This prospective study suggests that in symptomatic patients with rheumatic MS who are in sinus rhythm, the risk for stroke or systemic embolism is high.
- •
The presence of transient (<30 seconds), subclinical runs of AF detected on 24-hour Holter monitoring is a strong predictor of the occurrence of stroke or systemic embolism.
- •
The presence of subclinical AF on Holter monitoring in patients with MS in sinus rhythm may help in the risk stratification of these patients for stroke and systemic embolism.
Rheumatic heart disease (RHD) is a common cause of stroke in developing countries. Among patients with RHD, mitral stenosis (MS) is the valve lesion most strongly associated with stroke and systemic embolism. Pure or dominant MS was present in >90% of patients with stroke or systemic embolism in the widely cited series by Szekely. Embolism is more frequent in patients with MS and concomitant atrial fibrillation (AF), but about 1/5th of embolic episodes occur in patients who are in sinus rhythm (SR). There have been no randomized controlled studies in this population, but there is widespread consensus regarding the benefits of oral anticoagulation in patients with MS who are in AF or in those who have had embolic events in the past. However, the benefits of oral anticoagulation in patients with MS in SR are uncertain because of a paucity of data on the risk and predictors of embolism in these patients. Current recommendations for anticoagulation in this population are based on factors that are thought to increase the risk for stroke (such as an enlarged left atrium). Patients with MS in SR may frequently have transient, asymptomatic episodes of AF (subclinical AF) detected only on ambulatory electrocardiographic monitoring, which may potentially increase the risk for systemic embolism. We hypothesized that the occurrence of such episodes may be a better measure of the cumulative effect of numerous factors (such as patient age, stenosis severity, and left atrial [LA] size) which are thought to be related to the risk for systemic embolism. We therefore assessed the value of Holter-detected episodes of transient, subclinical AF in patients with MS in SR in predicting the occurrence of stroke or systemic embolism.
Methods
This was a single-center, prospective, observational study of patients with pure or dominant rheumatic MS in SR. Patients attending the cardiology outpatient clinics at a tertiary care teaching hospital were assessed for eligibility. Patients were approached for enrollment if they were ≥18 years of age and had rheumatic MS (mitral valve area [MVA] <2 cm 2 on echocardiography) as an isolated or dominant lesion and were in SR. In patients with multivalve disease, MS was considered dominant if it was the most severe of the lesions and was thought to be primarily responsible for the patients’ clinical symptoms. Patients were excluded if they had a history of stroke, transient ischemic attack (TIA), or non–central nervous system (CNS) systemic embolism or were receiving oral anticoagulation at enrollment. We also excluded patients who had any documented past episodes of AF. We did not include patients who had active infective endocarditis or were acutely ill, requiring hospitalization. In addition, only those eligible patients who consented to initial Holter examinations (which would have involved a few additional days stay for out-of-station patients) and were willing to undergo regular follow-up examinations were enrolled in the study. All patients gave written informed consent, and the study protocol was approved by the institutional ethics committee.
Detailed history was obtained from all patients, including information on the duration of disease, functional class, and history of any intervention or surgery for MS. We inquired about any history of stroke, TIA, or non-CNS systemic embolic events and looked for evidence of old CNS or peripheral embolic events on physical examination. A 12-lead electrocardiogram was obtained at study entry.
Transthoracic echocardiography (iE33; Philips Medical Systems, Bothell, Washington) was performed using standard protocols. MVA was estimated using continuous-wave Doppler by the pressure half-time method, supplemented by estimates obtained using planimetry. The severity of MS was graded per current recommendations (mild, >1.5 cm 2 ; moderate, 1 to 1.5 cm 2 ; and severe, <1.0 cm 2 ). LA dimensions were measured in the parasternal long-axis and the apical 4-chamber views, and LA volume was calculated using the prolate ellipse method. The LA appendage was visualized in the basal short-axis and 2-chamber views. The presence of thrombus in either of these chambers was diagnosed in the presence of an intracavitary echogenic mass distinct from the endocardium and seen in ≥2 views. Spontaneous echocardiographic contrast (SEC; dynamic smokelike echoes) in the left atrium was graded as mild or dense. We categorized SEC as mild if it was 1+ or 2+ and dense if it was 3+ or 4+, per the grading scheme suggested by Fatkin et al.
All patients underwent 24-hour electrocardiographic recording at study entry, using a 10-lead Holter system (Medilog Darwin FD12; Schiller, Belmont, Australia). Subclinical AF was diagnosed if the recording showed any asymptomatic episodes of definite AF. As in the study by Ramsdale et al, physicians interpreting the Holter recordings diagnosed AF in the presence of irregular tachycardia (>100 beats/min) in the presence of fibrillatory waves or in the absence of discernible p waves, irrespective of the duration of tachycardia. Episodes lasting ≥30 seconds were labeled as paroxysmal AF.
The primary outcome of interest was the composite of stroke, TIA, or non-CNS systemic embolism. Stroke was defined as any focal neurologic deficit that lasted >24 hours with or without brain imaging suggestive of a primary ischemic origin leading to tissue infarction. TIA was defined as physician-diagnosed focal neurologic deficit consistent with ischemia in a vascular territory, lasting <24 hours without tissue infarction. The diagnosis of non-CNS systemic embolism was made clinically (loss of arterial pulse or evidence of end-organ ischemia) and confirmed by imaging studies as appropriate. Patients were periodically contacted by a physician (RA) to inquire about the occurrence of outcomes and remind them regarding follow-up visits. Patients were encouraged to attend a follow-up visit at least once every 6 months, or earlier if an outcome had occurred. Outcomes were assessed by physicians who were unaware of the findings on Holter examination. In the event that patients were evaluated at other hospitals, we obtained details of clinical and imaging findings from patients and treating doctors when possible.
We expressed the occurrence of the composite primary outcome and its components as rates (events per patient-years) and presented them as Kaplan-Meier plots. We assessed the difference in the primary outcome between patients with and without subclinical AF using the log-rank test. We determined the independent predictive value of subclinical AF for the primary outcome using Cox regression with the following prognostically important variables in the model (model 1): duration of disease, severe functional class (New York Heart Association [NYHA] class III vs II), indexed MVA, indexed LA diameter, presence of dense SEC in the left atrium, thrombus in the LA appendage, and a history of intervention or surgery for mitral valve disease. We calculated adjusted hazard ratios (HRs) and their 95% confidence intervals (CIs) for each variable. We also constructed another Cox model by adding the following variables, whose prognostic importance for systemic embolism is less certain (model 2): male gender, presence of mitral regurgitation, and presence of aortic valve disease. We calculated the sensitivity, specificity, predictive values, and area under the receiver-operating characteristic curve of subclinical AF for predicting the primary outcome.
Mitral valve intervention (percutaneous or surgical) during the study period may potentially reduce the incidence of the primary outcome. But interventions are more likely to be performed in patients with severe disease, who are also more likely to experience primary outcome events. Therefore, we evaluated the protective effect of mitral intervention on the primary outcome after adjusting for the propensity to undergo intervention. We generated propensity scores for intervention using logistic regression, with duration of disease, indexed MVA, and severe functional impairment as predictor variables and intervention as the dependent variable. We generated HRs for mitral intervention in a Cox model, which included the propensity score and the variables from model 1 after excluding highly correlated (Pearson’s correlation coefficient >0.50) variables. A p value <0.05 was considered significant. All analyses were performed using Stata version 11 (StataCorp LP, College Station, Texas).
Results
From January to December 2012, we screened 256 patients with clinically diagnosed MS for possible enrollment. We excluded 23 patients after initial evaluation and echocardiography ( Figure 1 ). At the screening interview, 54 of the remaining 233 eligible patients indicated their inability or reluctance to attend follow-up visits or undergo baseline Holter examinations. The remaining 179 patients were enrolled in the present study. Patients were young (mean age 32 years) and had on average been symptomatic for nearly 6 years before presentation. Eighty percent of the patients had severe MS, and >50% were in NYHA class III. The left atrium was enlarged in all patients, with 99 (55%) having LA diameters ≥55 mm. Baseline characteristics are listed in Table 1 and details of other valve involvement in Table 2 . Transient, subclinical AF was detected on Holter monitoring in 49 patients (27%). None of the episodes lasted ≥30 seconds. No other atrial or ventricular arrhythmias were recorded.
| Variable | N = 179 |
|---|---|
| Age (years) | 32.6 ± 10.8 |
| Females | 106 (59%) |
| Known duration of disease (years) | 5.6 ± 2.8 |
| NYHA class | |
| I | 0 |
| II | 85 (47.5%) |
| III | 94 (52.5%) |
| IV | 0 |
| Indexed mitral valve area (cm 2 /m 2 ) ∗ | 0.57 ± 0.16 |
| Mitral valve area (cm 2 ) | |
| >1.5 | 5 (2.8%) |
| 1.0–1.5 | 32 (17.9%) |
| <1.0 | 142 (79.3%) |
| LA diameter (cm) † | 5.5 ± 0.9 |
| LA volume (cm 3 ) | 92.3 ± 41.9 |
| LAA thrombus | 15 (8.4%) |
| LA SEC | |
| None | 103 (57.5%) |
| Mild | 54 (30.2%) |
| Dense | 22 (12.3%) |
| RV systolic pressure (mm Hg) ‡ | 50.1 ± 22.6 |
| Previous mitral intervention | |
| Balloon valvuloplasty | 26 (14.5%) |
| Closed mitral commissurotomy | 6 (3.3%) |
∗ Absolute mitral valve area was 0.86 ± 0.24 cm 2 .
† In the parasternal long-axis view; indexed LA diameter was 3.7 (0.7) cm/m 2 .
‡ Estimated from the velocity of the tricuspid regurgitation jet.
| Lesion | Trivial | Mild | Moderate | Severe |
|---|---|---|---|---|
| Mitral regurgitation | 34 (19%) | 14 (7.8%) | 2 (1.1%) | 0 |
| Tricuspid regurgitation | 59 (33%) | 87 (48.6%) | 15 (8.4%) | 18 (10%) |
| Aortic stenosis | 2 (1.1%) | 1 (0.6%) | 0 | 0 |
| Aortic regurgitation | 80 (44.7%) | 35 (19.6%) | 4 (2.2%) | 0 |
At the end of the study period (July 2013), all patients had ≥1 final telephone or in-person follow-up visit. The average duration of follow-up was 10.2 ± 4 months. During the course of the study, 10 patients with LA appendage thrombi and 8 patients with SEC were prescribed oral anticoagulation by their treating physicians. At 18 months, 71 patients (40%) had undergone percutaneous mitral balloon valvuloplasty, and 4 had undergone mitral valve replacement. Eight patients had primary outcome events at the end of the study period, at a rate of 5.3 per 100 patient-years (95% CI 2.6 to 10.5). Four patients had strokes, 2 had TIAs, 1 had acute limb ischemia due to peripheral embolism, and 1 had central retinal artery occlusion. One of the patients with stroke and the patient with central retinal artery occlusion were managed at other hospitals. Events in these 2 patients were confirmed by typical history: sudden onset of hemiplegia in 1 patient and sudden monocular blindness in the other, which were evaluated by a physician (confirmed from patient discharge notes). The 6 remaining events were treated at our center and were confirmed on clinical examination and imaging. Of these 6 patients, 3 were in SR and 3 in AF at the time of the events.
Cumulative survival, free of the primary outcome, was significantly better in patients without compared with those with subclinical AF ( Figure 2 ). In univariate analysis, significant predictors of the primary outcome were the presence of subclinical AF on Holter monitoring (HR 4.54, 95% CI 1.08 to 19.0, p = 0.038) and the presence of dense SEC (HR 4.32, 95% CI 1.03 to 18.09, p = 0.045) ( Table 3 ).
