Planimetry of Mitral Valve Stenosis by Magnetic Resonance Imaging

Journal of the American College of Cardiology Vol. 45, No. 12, 2005 2005 by the American College of Cardiology Foundation ISSN 0735-1097/05/$30.00 Published by Elsevier Inc. doi:10.1016/j.jacc.2005.03.036 Planimetry of Mitral Valve Stenosis by Magnetic Resonance Imaging Behrus Djavidani, MD,* Kurt Debl, MD, Markus Lenhart, MD,* Johannes Seitz, MD,* Christian Paetzel, MD,* Franz X. Schmid, MD, Wolfgang R. Nitz, PHD,* Stefan Feuerbach, MD,* Günter Riegger, MD, Andreas Luchner, MD Regensburg, Germany OBJECTIVES BACKGROUND METHODS RESULTS CONCLUSIONS We sought to determine whether noninvasive planimetry of the mitral valve area (MVA) by magnetic resonance imaging (MRI) is feasible and reliable in patients with mitral stenosis (MS). Accurate assessment of MVA is particularly important for the management of patients with valvular stenosis. Current standard techniques for assessing the severity of MS include echocardiography (ECHO) and cardiac catheterization (CATH). In 22 patients with suspected or known MS, planimetry of MVA was performed with a 1.5-T magnetic resonance scanner using a breath-hold balanced gradient echo sequence (true FISP). Data were compared with echocardiographically determined MVA (ECHO-MVA, n 22), as well as with invasively calculated MVA by the Gorlin-formula at (CATH-MVA, n 17). The correlation between MRI- and CATH-MVA was 0.89 (p 0.0001), and the correlation between MRI- and ECHO-MVA was 0.81 (p 0.0001). The MRI-MVA slightly overestimated CATH-MVA by 5.0% (1.60 0.45 cm 2 vs. 1.52 0.49 cm 2,p NS) and ECHO-MVA by 8.1% (1.61 0.42 cm 2 vs. 1.48 0.42 cm 2,p 0.05). On receiveroperating characteristic curve analysis, a value of MRI-MVA below 1.65 cm 2 indicated mitral stenosis (CATH-MVA 1.5 cm 2 ), with a good sensitivity and specificity (89% and 75%, respectively). Magnetic resonance planimetry of the mitral valve orifice in mitral stenosis offers a reliable and safe method for noninvasive quantification of mitral stenosis. In the clinical management of patients with mitral stenosis, it has to be considered that planimetry by MRI slightly overestimates MVA, as compared with MVA calculated echocardiographically and at catheterization. (J Am Coll Cardiol 2005;45:2048 53) 2005 by the American College of Cardiology Foundation Cardiac magnetic resonance imaging (MRI) is a new method that allows one to evaluate cardiac structures and function noninvasively (1). Recently, the feasibility of velocity-encoded MRI, which closely correlates with echocardiography (ECHO), has been demonstrated for the assessment of stenotic aortic (2) and mitral valves (3). However, with new balanced gradient-echo sequences, direct visualization of valve area is also possible, and in patients with aortic stenosis, it has been shown that the valve area, as assessed by MRI planimetry, closely corresponds to the invasively assessed valve area, as assessed by cardiac catheterization (CATH) (4,5). Thus, planimetry of the mitral valve area (MVA) by MRI might as well offer an alternative noninvasive method in the diagnosis of mitral stenosis (MS). The aim of this study was to evaluate the accuracy and clinical utility of planimetry of the MVA by MRI in comparison with the echocardiographically calculated valve area (ECHO-MVA) and invasively derived MVA at catheterization (CATH-MVA). We hypothesized that planimetry of MVA by MRI is feasible and reliable and correlates From the Departments of *Diagnostic Radiology, Internal Medicine II, and Cardiothoracic Surgery, University Hospital of Regensburg, Regensburg, Germany. Drs. Djavidani and Debl contributed equally to the manuscript. Manuscript received October 7, 2004; revised manuscript received March 4, 2005, accepted March 10, 2005. well with ECHO- and CATH-MVA. To address this hypothesis, we evaluated 22 consecutive patients with suspected or known MS with all three methods. METHODS Patients. We studied 22 consecutive patients with suspected or known MS who were referred to our institution for further evaluation. All patients gave written, informed consent and tolerated the procedure well, and no adverse effects were observed. Patient characteristics are shown in Table 1. CATH. Left and right heart CATH was performed in 17 patients by an experienced operator. Four patients refused the invasive examination, and one patient was not examined at this time point, as he was in a reduced clinical condition (renal insufficiency, genitourinary infection, and moderate chronic obstructive pulmonary disease). Hemodynamic calculation of the MVA was performed using the Gorlin formula (6). In addition, biplane cine left ventriculography to calculate left ventricular function and coronary angiography to evaluate coronary arteries were performed. ECHO. The MVA was measured transthoracically by the Doppler pressure half-time method, as described previously (7). Evaluable data were available from all of the 22 patients. Each of these echocardiographic studies was performed by

JACC Vol. 45, No. 12, 2005 June 21, 2005:2048 53 Djavidani et al. MRI Planimetry of Mitral Stenosis 2049 Abbreviations and Acronyms CATH cardiac catheterization ECHO echocardiography MRI magnetic resonance imaging MS mitral stenosis MVA mitral valve area an investigator experienced in the technique, without knowledge of previous velocity or pressure gradient measurements. MRI studies. For left ventricular functional analysis, cine images were acquired on a 1.5-T scanner (Sonata, Siemens Medical Solutions, Erlangen, Germany) using a phased-array receiver coil and breath-hold acquisitions (13 heartbeats) prospectively gated to the electrocardiogram with a balanced GRE sequence (true FISP; slice thickness 8 mm, echo time 1.53 ms, pixel bandwidth 1,085 Hz, repetition time 3.14 ms leading to a temporal resolution of 43 ms, matrix 256 202). The number of Fourier lines per heartbeat was adjusted to allow the acquisition of 18 to 20 cardiac phases within a cardiac cycle. The field of view was 340 mm on average and adapted to the size of the patient. For the evaluation of the mitral valve, a retrospectively gated balanced GRE sequence (true FISP) approach was selected in order to cover the full cardiac cycle. The sequence was executed with a repetition time of 3.13 ms, and phase encoding was advanced every 1,200 ms. Data were normalized and resorted in time throughout the cardiac cycle, leading to a temporal resolution of 47 ms. The number of views per cardiac cycle was selected to be between 17 and 30, depending on the patient s heart rate. Data were collected during a breath-hold (14 heart beats, slice thickness 5 mm, echo time 1.57 ms, pixel bandwidth 930 Hz, matrix 256 205, excitation angle 59 ). The field of view was 340 mm times 276 mm on average and adapted to the size of the patient. Table 1. Patient Characteristics The imaging plane of the mitral valve was defined by acquiring diastolic four-chamber, three-chamber, and longaxis two-chamber views. The subsequent slices were defined perpendicular to the valvular plane and, in cases of orifices with an eccentric outlet, perpendicular to the origin of the jet in the left ventricle (Fig. 1). Depending on the morphology of the mitral valve, four to six slices parallel to the mitral valve orifice (short-axis view) were assessed. Starting at the base of the valve and moving toward the tip of the cusps, the visible orifice area continuously decreased. The most apical slice that was still located at the maximum diastolic opening on the tip of the cusps was chosen for planimetry of MVA. We selected the time point within the cardiac cycle at which maximum diastolic mitral valve opening was observed in the three-chamber view. At precisely the same time point, MVA was planimetered in the shortaxis view. We placed our traces at the point of the bright pixels. Areas of signal void in severely calcified valves or in areas of very turbulent flow were counted as part of the valve leaflet. Planimetry was performed by two independent observers who were unaware of each other s interpretations and the ECHO and CATH results. At least three measurements were performed by each observer. For calculating MVA, the mean value of the two observers was used. Image quality was semiquantitatively scored by each observer in three grades (good, moderate but still evaluable, and not sufficient for evaluation). Statistical analysis. Results are expressed as the mean value SD. The agreement among the three methods of quantification of the MVA was assessed by univariate regression analysis and by the Bland-Altman method. Differences in mean values between two groups were analyzed by the Student t test. The chi-square test was performed to compare frequencies between groups. The ROC analysis was carried out to determine predictive values of MRI to detect MS (MVA 1.5 cm 2 ), as defined by catheterization. MRI-MVA >1.5 cm 2 (n 14) MRI-MVA <1.5 cm 2 (n 8) Age (yrs) 61 4 62 4 Gender (% female) 86 50 CAD (%) 0 13 NYHA functional class (%) I 15 0 II 64 37 III 21 63* Prior valvuloplasty (%) 43 13 Sinus rhythm (%) 50 62 Mitral regurgitation (%) 0 I 64 100 II III 36 0 MRI-CO (l/min) 4.1 0.3 3.6 0.4 MRI-EF (%) 59 2 59 4 *p 0.05. Mean SEM. CAD coronary artery disease; CO cardiac output; EF ejection fraction; MRI magnetic resonance imaging; MVA mitral valve area; NYHA New York Heart Association.

2050 Djavidani et al. JACC Vol. 45, No. 12, 2005 MRI Planimetry of Mitral Stenosis June 21, 2005:2048 53 Figure 1. Magnetic resonance images to define the imaging plane. Diastolic two-chamber (a), four-chamber (b), and three-chamber (c) views. A jet originating from the mitral valve leaflets is visible. The subsequent slices were defined perpendicular to the valvular plane and, in cases of orifices with an eccentric outlet, perpendicular to the origin of the jet in the left ventricle. Visualization of the diastolic opening of the mitral valve orifice. (d) Systolic image parallel to the closed mitral valve. (e) Diastolic image of the same plane showing the small orifice in this patient. (f) Planimetry as drawn in with a resulting orifice area of 0.6 cm 2, representing severe mitral stenosis (corresponding cardiac catheterization-mitral valve area resulted also in 0.6 cm 2 and echocardiography-mitral valve area resulted in 0.7 cm 2 ). A level of significance of 0.05 was defined as statistically significant. RESULTS Patient characteristics. Most patients were symptomatic with exertional dyspnea, and New York Heart Association functional class III was more prevalent in patients with more severe MS (p 0.05). Patients with less severe MS have had previous valvuloplasty more frequently (p NS) and more concomitant mitral regurgitation (p NS). Cardiac output (derived from cardiac MRI) was higher in patients with less severe MS (p NS). Patient characteristics are depicted in Table 1. MVA. The mean MRI-MVA determined by planimetry was 1.61 0.42 cm 2 (range 0.6 to 2.50 cm 2 ). Hemodynamic results (i.e., cardiac output and ejection fraction) are shown in Table 1. Image quality in patients with sinus rhythm (n 12) was of overall good quality. Image quality in patients with atrial fibrillation was of good quality in 6 and of moderate quality in 4 of 10 patients. Interobserver and intraobserver variabilities (coefficient of variation) were 0.03 0.01 cm 2 and 0.04 0.02 cm 2, respectively. As assessed by ECHO, the mean ECHO-MVA was 1.48 0.42 cm 2 (range 0.70 to 2.50 cm 2 ), and as assessed by cardiac catheterization, the mean CATH-MVA calculated by the Gorlin-formula was 1.52 0.49 cm 2 (range 0.60 to 2.4 cm 2 ). Comparison of MRI with CATH. The correlation between MRI- and CATH-MVA was very good (n 17, r 0.89, p 0.0001). The mean absolute difference between MVA derived by MRI and the Gorlin formula was 0.08 0.22 cm 2 (p NS), resulting in a slight but not significant overestimation of the MRI-MVA compared with CATH- MVA by 5.0%. As shown in the Bland-Altman analysis, there were no outliers beyond 2 SD of the mean difference of both methods (Fig. 2). Comparison of MRI with ECHO. The correlation between planimetry of the MRI- and ECHO-MVA was very good (r 0.81, p 0.0001). The mean absolute difference between MVA derived by MRI and ECHO was 0.13 0.24 cm 2 (p 0.05), resulting in a slight overestimation of the MRI-MVA as compared with ECHO-MVA by 8.1% (Fig. 3). Comparison of ECHO to CATH. The correlation between ECHO-MVA and CATH-MVA was close (n 17, r 0.69, p 0.002). The mean absolute difference between MVA derived by ECHO and CATH was 0.04 0.29 cm 2 (p NS) (Fig. 4). Predictive values of MRI in detecting stenosis. As assessed by ROC analysis, MRI detected MS (defined as

JACC Vol. 45, No. 12, 2005 June 21, 2005:2048 53 Djavidani et al. MRI Planimetry of Mitral Stenosis 2051 Figure 2. (A) Scattergram of mitral valve area (MVA) determined by magnetic resonance imaging (MRI) and cardiac catheterization (CATH) in 17 patients. (B) Bland-Altman plot of the average mean versus the differences between MRI planimetry and MVA derived at CATH. The solid line is the mean difference; the dotted lines mark the standard deviations of the differences. MVA 1.5 cm 2 at catheterization) at a MRI-MVA threshold of 1.65 cm 2 with high accuracy (sensitivity 89% and specificity 75%, ROC area 0.9). The positive and negative predictive values were 80% and 86%, respectively (Table 2). DISCUSSION The current study demonstrates that visualization and planimetry of MVA by MRI is possible and reliable in patients with MS. We found a very good correlation of MVA planimetry with the invasively derived data by the Gorlin formula and the noninvasively derived data by the Doppler pressure half-time method. Furthermore, MRI obtained excellent predictive values for the detection of MS. In patients with MS, the symptoms of dyspnea and fatigue are often difficult to quantify and do not always reflect genuine reductions in mitral orifice area (8). Because of the continuous and progressive nature of MS, repeated assessment of the valve area is often necessary, and the ability to accurately assess the valve area noninvasively is of great importance. In this respect, the excellent predictive values suggest that MRI as a noninvasive method might prove highly useful for the stratification of patients with Figure 3. (A) Scattergram of MVA determined by MRI and echocardiography (ECHO) in 22 patients. (B) Bland-Altman plot of the average mean versus the differences between MRI planimetry and MVA derived at ECHO. The solid line is the mean difference; the dotted lines mark the standard deviations of the differences. Abbreviations as in Figure 2.

2052 Djavidani et al. JACC Vol. 45, No. 12, 2005 MRI Planimetry of Mitral Stenosis June 21, 2005:2048 53 Figure 4. (A) Scattergram of MVA determined by ECHO and CATH in 17 patients. (B) Bland-Altman plot of the average mean versus the differences between MRI planimetry and MVA derived at CATH. The solid line is the mean difference; the dotted lines mark standard deviations of the differences. Abbreviations as in Figures 2 and 3. suspected MS. In addition to the assessment of valve area, MRI is also able to detect leaflet or chordal apparatus thickening and left atrial or atrial appendage thrombi and allows one to assess the presence and severity of concomitant regurgitation. Consideration of the valve area, as well as these important aspects, is of pivotal importance for therapeutic decision-making and either conservative, percutaneous, or surgical management of patients with MS. In the current study, we report a slightly higher mean MVA by MRI with the methods used in this study, in comparison to CATH ( 5 %) and ECHO ( 8 %). In addition, ROC analysis determined that detection of CATH-MVA 1.5 cm 2 was best achieved at a MRI-MVA threshold of 1.65 cm 2. Both findings therefore demonstrate a slight but systematic overestimation of MVA by MRI. The underlying mechanism is currently unclear and may be related to MRI technology or the direct visualization of valve opening, as compared with indirect estimation by CATH or ECHO. Another explanation for the overestimation of valve orifice might be volume averaging from the cine MRI slice thickness chosen. Transplanar valve motion during diastole might also lead to overestimation of the valve area, particularly when the imaging plane misses the smallest orifice area. Nevertheless, we addressed this problem by acquiring at least four slices at different levels of the mitral valve and recommend this approach to minimize the potential of mitral valve area overestimation because of imprecise localization. Nevertheless, the obtained cut-off values, which have been calculated retrospectively, will now have to be assessed in a prospective study. In routine clinical practice, both CATH and ECHO are used to assess MS. Cardiac catheterization is widely accepted as the gold standard, and MVA is calculated by the Gorlin formula from the mean transvalvular gradient and valvular flow (6). Because of the indirect measurements, however, the accuracy of MVA calculation has often been challenged (9,10). Despite these limitations, the Gorlin formula remains the standard criterion against which new noninvasive methods must be judged. Visualization and planimetry of MVA by ECHO provide a noninvasive measurement and are widely used for estimation of MVA (11 13). Although the Doppler pressure half-time method used in this study has proven valuable in the noninvasive assessment of the severity of MS in a variety of circumstances, its accuracy may also be limited in a variety of conditions (7,14). Together, this variability in MVA measurements by different methods underscores the importance of understanding the assumptions and limitations of each method and supports a noninvasive method for direct determination of MVA by planimetry such as MRI. The current study is the first to utilize cine MRI for direct visualization and planimetry of MVA, and our results indicate that MRI planimetry is accurate. This new method therefore comple- Table 2. MRI Planimetry of MS, Predictive Values for Given CATH-MVA CATH-MVA (cm 2 ) Cases/Total (n) MRI-Cutoff (cm 2 ) ROC Area (95% CI) Sens./Spec. (%) PPV/NPV (%) MVA 1.5 cm 2 10/17 1.65 0.90 (0.76 1.0) 89/75 80/86 CATH cardiac catheterization; CI confidence interval; MRI-Cutoff optimal cutoff for MRI-MVA to detect the given CATH-MVA; MS mitral stenosis; MVA mitral valve area; NPV negative predictive value; PPV positive predictive value; ROC receiver-operator characteristic; Sens. sensitivity; Spec. specificity.

JACC Vol. 45, No. 12, 2005 June 21, 2005:2048 53 Djavidani et al. MRI Planimetry of Mitral Stenosis 2053 ments established MRI methods of valve assessment, such as velocity mapping (2,3). Conclusions. Magnetic resonance imaging allows for noninvasive planimetry of MVA with good image quality and also provides additional information on valve structure and function. In the clinical management of patients with MS, it has to be considered that, despite a close correlation, MRI with the methods used in this study slightly overestimates MVA, as compared with CATH and ECHO. Reprint requests and correspondence: Dr. Behrus Djavidani, Department of Radiology, University Hospital of Regensburg, Franz-Josef-Strauss-Allee 11, D-93042 Regensburg, Germany. E-mail: behrus.djavidani@klinik.uni-regensburg.de. REFERENCES 1. Earls JP, Ho VB, Foo TK, Castello E, Flamm SD. Cardiac MRI: recent progress and continued challenges. J Magn Reson Imaging 2002;16:111 27. 2. Caruthers SD, Lin SJ, Brown P, et al. Practical value of cardiac magnetic resonance imaging for clinical quantification of aortic valve stenosis: comparison with echocardiography. Circulation 2003;108: 2236 43. 3. Lin SJ, Brown PA, Watkins MP, et al. Quantification of stenotic mitral valve area with magnetic resonance imaging and comparison with Doppler ultrasound. J Am Coll Cardiol 2004;44:133 7. 4. Friedrich MG, Schulz-Menger J, Poetsch T, et al. Quantification of valvular aortic stenosis by magnetic resonance imaging. Am Heart J 2002;144:329 34. 5. John AS, Dill T, Brandt RR, et al. Magnetic resonance to assess the aortic valve area in aortic stenosis. J Am Coll Cardiol 2003;42: 519 26. 6. Gorlin R, Gorlin SG. Hydraulic formula for calculation of the area of the stenotic mitral valve, other cardiac valves and central circulatory shunts. Am Heart J 1951;41:1 29. 7. Hatle L, Angelsen B, Tromsdal A. Noninvasive assessment of atrioventricular pressure half-time by Doppler ultrasound. Circulation 1979;60:1096 104. 8. Hugenholtz PG, Ryan TJ, Stein SW, et al. The spectrum of pure mitral stenosis: hemodynamic studies in relation to clinical disability. Am J Cardiol 1962;10:773. 9. Bryg RJ, Williams GA, Labovitz AJ, et al. Effect of atrial fibrillation and mitral regurgitation on calculated MVA in mitral stenosis. Am J Cardiol 1986;57:634 8. 10. Axler O, Tousignant C, Thompson CR, et al. Comparison of transesophageal echocardiography, Fick, and thermodilution in cardiac output in critically ill patients. J Crit Care 1996;11:109 16. 11. Nichol PM, Gilbert BW, Kisslo JA. Two-dimensional echocardiographic assessment of mitral stenosis. Circulation 1977;55:120 8. 12. Wann LS, Weyman AE, Feigenbaum H, et al. Determination of mitral valve area by cross-sectional echocardiography. Ann Intern Med 1978;88:337 41. 13. Henry WL, Griffith J, Michaelis LL, et al. Measurement of mitral orifice area in patients with mitral valve disease by real-time twodimensional echocardiography. Circulation 1975;51:827 31. 14. Nakatani S, Masuyama T, Kodama K, et al. Value and limitations of Doppler echocardiography in the quantification of stenotic mitral valve area: comparison of the pressure half-time and the continuity equation methods. Circulation 1988;77:78 85.