In patients with mitral stenosis (MS) in sinus rhythm (SR), guidelines recommend anticoagulation if the left atrium is enlarged based on diameter measurements. We sought to compare the association of left atrial (LA) diameter and LA volume with markers of thromboembolic risk (peak LA appendage emptying velocity [LAAv] and LA spontaneous contrast density) measured during transesophageal echocardiography in 152 patients with moderate to severe MS. High thromboembolic risk was defined by a peak LAAv <25 cm/s and/or dense spontaneous contrast. Mean LA diameter (50 ± 7 mm, 32 to 77) and LA volume (152 ± 70 ml, 67 to 720) were significantly correlated (r = 0.71, p <0.0001), but the relation was curvilinear and the 95% confidence interval increased with LA diameter. In the subset of 80 patients in SR who underwent clinically indicated transesophageal echocardiography, body surface area (BSA)-indexed LA volume but not LA diameter differentiated patients with normal from those with low LAAv (86 ± 17 vs 71 ± 17 ml/m 2 , p <0.01, and 50 ± 6 vs 48 ± 6 mm, p = 0.13, respectively) and patients with dense spontaneous contrast from those with no or mild spontaneous contrast (81 ± 16 vs 63 ± 15 ml/m 2 , p <0.01, and 49 ± 6 vs 46 ± 5 mm, p = 0.11, respectively). BSA-indexed LA volume provided the highest area under the curve (0.85) for high thromboembolic risk and LA diameter the lowest (0.65). A BSA-indexed LA volume >60 ml/m 2 provided an excellent 90% sensitivity despite 44% specificity, 76% positive predictive value, and 70% negative predictive value. Use of this threshold instead of 50 or 55 mm would have changed the indication for anticoagulation in 51% to 77% of patients. In conclusion, LA volume was more strongly associated with markers of thromboembolic risk than LA diameter, which poorly reflected LA size. Our results support the use of BSA-indexed LA volume to guide the decision for anticoagulation in patients with MS in SR, which may lead to significant change in the management of those patients. We suggest a threshold of 60 ml/m 2 , which has good sensitivity, albeit with low specificity.
Peak left atrial appendage emptying velocity (LAAv) and LA spontaneous contrast density measured during transesophageal echocardiography are well-validated markers of thromboembolic risk. The aim of the present study was to compare LA diameter and LA volume assessment of LA remodeling and their respective association to these markers of thromboembolic risk in patients with mitral stenosis (MS) in sinus rhythm (SR).
Methods
Patients with MS who underwent transthoracic echocardiography performed by the last author (DM-Z) were prospectively enrolled in the present study. Inclusion criteria were moderate or severe MS (defined as mitral valve area <1.5 cm 2 ) and absence of mitral regurgitation of grade >2. Of the total population, 80 patients also underwent clinically indicated transesophageal echocardiography to exclude LA thrombus (before percutaneous mitral commissurotomy or surgery) and were in SR. All echocardiograms were clinically indicated and verbal consent was obtained. Because this was a nontherapeutic study and patients required no additional investigations, the institutional research ethics committee deemed that formal ethical permission was not required.
All transthoracic and transesophageal echocardiographies were performed by the last operator on the same day using state-of-the-art commercially available systems. LA diameter was measured as recommended in the parasternal long-axis view. LA volume was calculated using the biplane area–length method in 4- and 3-chamber views as previously validated. LAA function was assessed using transesophageal echocardiographic peak emptying velocity measured by placing the sample volume of the pulse-wave Doppler in the LAA cavity. Three to 5 consecutive cycles were averaged. Decreased LAA function was defined by a peak LAAv <25 cm/s. LA spontaneous contrast was diagnosed using transesophageal echocardiographic presence of dynamic smokelike echoes within the left atrium or LAA with a characteristic swirling motion that was distinct from white noise artifact. Severity of LA spontaneous contrast was graded from 0 to 4 (0 = none, i.e., absence of echogenicity; 1 = mild, i.e., minimal echogenicity detectable only transiently during the cardiac cycle with optimal gain settings; 2 = mild to moderate, i.e., transient spontaneous echocardiographic contrast without increased gain settings and more dense pattern than grade 1; 3 = moderate, i.e., dense swirling pattern during entire cardiac cycle; 4 = severe, i.e., intense echodensity and very slow swirling patterns in LAA, usually with a similar density in the main LA cavity) as previously reported. Dense spontaneous contrast corresponded to grade 2 to 4. Assessment of MS severity was based on mitral valve area measured using the planimetry method and mean transmitral gradient. Mitral regurgitation was semiquantitatively graded from 0 to 4. Systolic pulmonary artery pressure was assessed using continuous-wave Doppler. Left ventricular dimension and function were measured as recommended by the American Society of Echocardiography.
Data are presented as mean ± SD. Correlation between LA diameter and LA volume was studied using linear and nonlinear regressions and the 95% confidence interval of estimated LA volume was used to assess closeness of fit and its trends for change with increasing M-mode diameter. A t test was used to compare LA size between groups. The predictive value of LA diameter and LA volume for thromboembolic risk (peak LAAv <25 cm/s, dense spontaneous contrast, or peak LAAv <25 cm/s and dense spontaneous contrast) was analyzed. Sensitivity, specificity, and positive and negative predictive values were determined for various thresholds of LA size and area under the curve of receiver operating characteristic analysis was calculated. A p value <0.05 was considered statistically significant.
Results
Population characteristics and subgroup descriptions are presented in Table 1 . In the total cohort (152 patients), mean LA diameter and LA volume showed significant correlation (r = 0.71, p <0.0001; Figure 1 ). However, the relation was curvilinear and the 95% confidence interval increased with LA diameter (75 to 158 ml for 40 mm, 92 to 248 ml for 50 mm, and 117 to 620 ml for 60 mm).
| Variable | Total Study Population | Subgroup in SR Who Underwent TEE |
|---|---|---|
| (n = 152) | (n = 80) | |
| Men | 41 (27%) | 20 (25%) |
| Age (years) | 53 ± 14 | 52 ± 16 |
| New York Heart Association class III/IV | 74 (49%) | 34 (43%) |
| Mitral valve area (cm 2 ) | 1.1 ± 0.2 | 1.1 ± 0.3 |
| Mean transmitral gradient (mm Hg) | 10 ± 5 | 10 ± 6 |
| Left atrial diameter (mm) | 50 ± 7 | 49 ± 6 |
| Left atrial diameter/body surface area (mm/m 2 ) | 29 ± 4 | 28 ± 4 |
| Left atrial volume (ml) | 152 ± 70 | 135 ± 34 |
| Left atrial volume/body surface area (ml/m 2 ) | 87 ± 37 | 77 ± 18 |
| Left ventricular end-diastolic diameter (mm) | 50 ± 6 | 50 ± 7 |
| Left ventricular end-systolic diameter (mm) | 31 ± 6 | 31 ± 7 |
| Ejection fraction (%) | 66 ± 10 | 66 ± 9 |
In the subset of 80 patients in SR who underwent transesophageal echocardiography, 10 (16%) had a history of an embolic event and 33 (42%) were on anticoagulant therapy. Mean LAAv was 30 ± 13 cm/s (12 to 70) and 35 patients (44%) had a low LAAv (<25 cm/s). No patients had LA thrombus. As presented in Table 2 , LA diameter was not significantly different between patients with normal and low LAAv (p = 0.13). In contrast, body surface area (BSA)-indexed LA diameter (p <0.05) and, more significantly, LA volume (absolute and BSA indexed, p <0.01) were larger in patients with low LAAv than in patients with normal LAAv. Area under the curve was only 0.62 for LA diameter and increased to 0.73 for BSA-indexed LA volume. LA diameter >50 mm produced 60% sensitivity, 69% specificity, 60% positive predictive value, and 69% negative predictive value for identifying low LAAv and a diameter of 55 mm produced respective values of 20%, 84%, 50%, and 58%. In contrast, BSA-indexed LA volume >60 ml/m 2 produced values of 91%, 27%, 49%, and 80%, and a higher threshold of 80 ml/m 2 produced values of 69%, 71%, 65%, and 74% respectively.
| Variable | LAAv (n = 80) | p Value | AUC | |
|---|---|---|---|---|
| Low | Normal | |||
| Left atrial diameter (mm) | 50 ± 6 | 48 ± 6 | 0.13 | 0.62 |
| Left atrial diameter/body surface area (mm/m 2 ) | 29 ± 3 | 27 ± 4 | <0.05 | 0.64 |
| Left atrial volume (ml) | 149 ± 35 | 124 ± 30 | <0.01 | 0.71 |
| Left atrial volume/body surface area (ml/m 2 ) | 86 ± 17 | 71 ± 17 | <0.01 | 0.73 |
| SEC (n = 47) | ||||
|---|---|---|---|---|
| Dense | No or Mild | |||
| Left atrial diameter (mm) | 49 ± 6 | 46 ± 5 | 0.11 | 0.65 |
| Left atrial diameter/body surface area (mm/m 2 ) | 29 ± 3 | 27 ± 4 | <0.05 | 0.67 |
| Left atrial volume (ml) | 138 ± 28 | 110 ± 22 | <0.01 | 0.78 |
| Left atrial volume/body surface area (ml/m 2 ) | 81 ± 16 | 63 ± 15 | <0.01 | 0.80 |
| LAAv + SEC (n = 47) | ||||
|---|---|---|---|---|
| High Risk | Low Risk | |||
| Left atrial diameter (mm) | 49 ± 6 | 46 ± 6 | 0.10 | 0.65 |
| Left atrial diameter/body surface area (mm/m 2 ) | 29 ± 3 | 26 ± 4 | <0.01 | 0.73 |
| Left atrial volume (ml) | 137 ± 28 | 108 ± 20 | <0.01 | 0.81 |
| Left atrial volume /body surface area (ml/m 2 ) | 81 ± 16 | 61 ± 12 | <0.01 | 0.85 |
Because anticoagulation can decrease spontaneous contrast density and because there was an interaction between LA size and anticoagulation therapy (65% of patients in highest LA volume quartile were on anticoagulation therapy vs 33% in 3 other quartiles, p <0.05), the 33 patients taking an anticoagulant were excluded from the present analysis. In the remaining 47 patients, LA diameter was not significantly different between patients with dense spontaneous contrast and those with no or mild spontaneous contrast (p = 0.11), whereas BSA-indexed LA diameter (p <0.05) and LA volume (absolute and BSA indexed, p <0.01) were significantly larger ( Table 2 ). BSA-indexed LA volume produced the highest area under the curve for dense spontaneous contrast (0.80) and LA diameter the lowest (0.65). LA diameter >50 mm produced 45% sensitivity, 67% specificity, 68% positive predictive value, and 43% negative predictive value for identifying dense spontaneous contrast, and LA diameter >55 mm produced 10%, 89%, 60%, and 38% respectively. In contrast, a BSA-indexed LA volume >60 ml/m 2 was much more sensitive, with 90% sensitivity, 28% specificity, 67% positive predictive value, and 63% negative predictive value, whereas a threshold of 80 ml/m 2 produced 59%, 89%, 89% and 57%, respectively.
Of the 47 patients in SR not taking an anticoagulant, 31 (66%) had a low LAAv (<25 cm/s) or dense spontaneous contrast. There was no significant difference in LA diameter between patients with low or high markers of thromboembolic risk (p = 0.10), whereas BSA-indexed LA diameter and LA volume (nonindexed and BSA indexed) were significantly larger in patients with high markers of thromboembolic risk ( Table 2 ). BSA-indexed LA volume produced the highest area under the curve for high thromboembolic risk (0.85; Figure 2 ). An LA diameter threshold of 50 mm had poor sensitivity (42%) and average specificity (63%). A threshold of 55 mm produced good specificity (89%) but very poor sensitivity (10%). In contrast, BSA-indexed LA volume >60 ml/m 2 produced excellent 90% sensitivity despite 44% specificity, 76% positive predictive value, and 70% negative predictive value and a threshold of 80 ml/m 2 produced 58%, 94%, 95%, and 54%, respectively.
