Assessment of left ventricular mass regression after aortic valve replacement cardiovascular magnetic resonance versus M-mode echocardiography

Transcription

1 European Journal of Cardio-thoracic Surgery 24 (2003) Abstract Assessment of left ventricular mass regression after aortic valve replacement cardiovascular magnetic resonance versus M-mode echocardiography Kim Rajappan a,c, *, Nicholas G. Bellenger a, Giovanni Melina b, Marco Di Terlizzi a, Magdi H. Yacoub b, Desmond J. Sheridan c, Dudley J. Pennell a a Cardiovascular MR Unit, Royal Brompton Hospital, Sydney Street, London, UK b Academic Department of Cardiothoracic Surgery, Royal Brompton and Harefield NHS Trust, Sydney Street, London, UK c Academic Cardiology Unit, St Mary s Hospital, NHLI, Imperial College School of Medicine, London, UK Received 8 February 2003; received in revised form 20 March 2003; accepted 21 March 2003 Objective: In patients with aortic valve disease, the presence of left ventricular hypertrophy (LVH) carries a significant risk of adverse cardiovascular events. Regression of hypertrophy after aortic valve replacement (AVR) is associated with a reduction in risk. In general, M-mode echocardiography has been used for quantitative assessment of left ventricular mass (LVM) and regression, but this technique is believed to have limitations from which cardiovascular magnetic resonance (CMR) does not suffer. The objective of this study therefore was to determine whether quantitative assessment of LVM and regression after AVR using the two techniques was comparable. Methods: Thirty-nine patients with aortic valve disease were studied before and 1 year after AVR. Transthoracic M-mode echocardiography and four different formulae were used to calculate left ventricular mass index (LVMI), and then compared with CMR measurements. Results: Overall, correlation between the techniques for single measurement of LVMI was moderate (r-values from 0.64 to 0.69), with a tendency for overestimation by echocardiography; there was no agreement in degree of regression (r-values from to 0.18). The Bland Altman limits of agreement ranged from 85 to 131% for single measurement of LVMI, and % for regression. The change in LVMI with CMR was 43 ^ 28 g/m 2, vs. 27 to 54 ^ 19 to 41 g/m 2 using echocardiography. Conclusions: M-mode echocardiography does not provide reliable quantification of regression of LVH in individuals, and for accurate measurement CMR is superior. The use of CMR in future studies may reduce costs since fewer subjects are needed to accurately detect significant changes in LVMI after AVR. q 2003 Elsevier Science B.V. All rights reserved. Keywords: Valve replacement; Hypertrophy; Regression; Echocardiography; Cardiovascular magnetic resonance 1. Introduction Left ventricular hypertrophy (LVH) in patients with aortic valve disease may initially be seen as an attempt to normalise wall stress and oxygen demand. However, as with hypertensive LVH, it is associated with unfavourable alterations in myocardial pathophysiology i.e. left ventricle (LV) systolic dysfunction, ischaemia, and arrhythmias, as well as adverse prognosis [1]. Regression of LVH has been demonstrated extensively after aortic valve replacement (AVR) [2], but the persistence of LVH is associated with a * Corresponding author. Tel.: þ ; fax: þ address: kumaran.rajappan@ic.ac.uk (K. Rajappan). worse outcome [3], with patient-related factors, particularly systemic blood pressure, being significant causes of late residual LVH [4]. This has focused attention on the ability to achieve regression of LVH, and how best to measure the change. To date, most studies of LVH and its regression after AVR have used M-mode transthoracic echocardiography, a one-dimensional (1D) technique. This is readily available and the geometrical assumptions used have been validated in normal hearts [5], but it has some significant limitations, including being operator dependant, acoustic window dependant and subject to significant inter- and intraobserver variability. With the advent of three-dimensional (3D) tomographic techniques such as 3D echocardiography and ultrafast computerised tomography, it is now possible to /03/$ - see front matter q 2003 Elsevier Science B.V. All rights reserved. doi: /s (03)

2 60 K. Rajappan et al. / European Journal of Cardio-thoracic Surgery 24 (2003) address many of these limitations, but with cardiovascular magnetic resonance (CMR), it is possible to obtain highresolution images in any desired plane without the need for ionising radiation or limitation by acoustic windows [6]. A stack of contiguous short axis slices that encompass the entire LV can be acquired and the precise volumes, mass and function calculated without the need for geometric assumptions. This results in measurements that are not only accurate compared to anatomical LVM determined by autopsy [7] but highly reproducible [8] and reduces the patient numbers required to prove a hypothesis in research studies [9]. Furthermore, the current fast sequences allow this to be achieved in a shorter imaging time than many of the other techniques [10]. Previous work from our institution has compared various techniques for measurement of cardiac volumes and function [11]. A few studies have compared echocardiography and CMR for assessment of LVM in various patient populations [12,13] but to date none have looked at quantitative assessment of regression of LVM. The aim of this work was therefore to determine how conventional 1D echocardiography compares with CMR, for the assessment of regression of LVM in a group of patients undergoing valve replacement surgery for aortic valve disease. 2. Materials and methods 2.1. Study population Thirty-nine patients (30 males) undergoing AVR were recruited to this study (aged 67 ^ 10 years), and 95% of them had predominant AS. The only exclusion criteria were those that prevented CMR. Of the 39 patients who had preoperative investigations, 28 underwent both echocardiography and CMR 1 year post-operatively. Of the 11 patients who did not undergo CMR post-operatively four patients had permanent pacemaker implantation, one became claustrophobic, one had an epicardial pacing wire left in situ, and five were unable to attend/lost to follow-up. The Research and Ethics Committee approved the study, and patients gave informed written consent. Procedures were carried out in accordance with local guidelines Surgical technique Twenty-nine valves were inserted in a standard manner. Eleven bioprostheses (five Baxter Carpentier Edwards, four Baxter Edwards Perimount, one Mitroflow and one St Jude) and 18 mechanical valves (nine Carbomedics 500 Type A, five St Jude, three Baxter Edwards MIRA and one Medtronic Hall) were used. Ten prostheses (three Medtronic Inc. Freestyle xenograft and seven homograft) were inserted as an aortic root replacement with reimplantation of the coronary arteries [14] Echocardiography Transthoracic echocardiography was performed using an ATL HDI 3000 (ATL Ltd, Bothwell, Washington, USA) cardiac ultrasound scanner with a 2.5 MHz probe, according to American Society of Echocardiography (ASE) guidelines [15]. LV dimensions were measured at end-diastole by 2D guided M-mode echocardiography taken at the level of the papillary muscle tip. LVM was calculated using the equations below and corrected for body surface area to give LVM index (LVMI). All echocardiographic values were the mean of three separate readings. Bennett Evans [16]: LVM ¼ [(2 LVW þ LVID) 3 2 LVID 3 ] 1.05 ASE [17]: LVM ¼ 0.83 [(IVST þ PWT þ LVID) 3 2 LVID 3 ] Penn [5]: LVM ¼ 1.04 [(IVST þ PWT þ LVID) 3 2 LVID 3 ] Teichholz [18]: LVM ¼ 1.04 {[7/(2.4 þ LVID þ IVST þ PWT)] (LVID þ IVST þ PWT) 3 2 [7/(2.4 þ LVID)] (LVID) 3 } where LVW is the left ventricular wall thickness measured from the posterior wall, LVID the left ventricular internal diameter, IVST the interventricular septal thickness, PWT is the posterior wall thickness. Established criteria were used to define the presence of LVH (LVMI $ 134 or 110 g/m 2 in men and women, respectively) [19] Cardiovascular magnetic resonance (CMR) Patients were imaged with a Picker Edge 1.5T scanner (Picker, Cleveland, OH), using the body coil and electrocardiogram (ECG) triggering, as previously described [20]. In brief, the cardiac short axis was determined from three scout images of the LV, the transverse, vertical long axis and breath-hold end-diastolic horizontal long axis. The basal short axis slice was positioned just forward of the atrioventricular ring, and all subsequent breath-hold cines were acquired in 1 cm steps towards the apex. A breath-hold segmented gradient echo fast low-angle shot (FLASH) sequence was used for each of the contiguous short axis slices. Parameters were as follows: echo time (TE) 3.8 ms, repeat time (TR) ¼ RR interval, slice thickness 10 mm, field of view cm, read matrix 256, phase matrix 128, frames 16, flip angle 358, phase encode group An average of 11 short axis segments was needed to encompass the entire LV. Image analysis was performed on a personal computer using in-house developed software (CMRtools q Imperial College). End-diastolic (phase 1) and end-systolic (phase 4 7 depending on number of phase encoded groups) images

3 K. Rajappan et al. / European Journal of Cardio-thoracic Surgery 24 (2003) were chosen as the maximal and minimal cross-sectional areas in each cine. Short axis end-diastolic epicardial and endocardial borders were traced manually for each slice (Fig. 1). From the area within the contours and the slice thickness, the epicardial and endocardial volumes were calculated, the difference representing myocardial volume. Mass was derived from this volume multiplied by the specific density of myocardium (1.05 g/cm 3 ). Papillary muscles were included in the mass. Care was taken not to include atrial slices at end-systole secondary to apical movement of the heart Statistical analysis The mean differences in LVM and LVMI were calculated between each technique and Student s t-test was used to determine statistical significance. A line of identity on a scattergram was drawn between each technique to allow a visual assessment of agreement. The correlation coefficient was then calculated to assess the strength of the relation. However, since correlation does not necessarily represent agreement, Bland Altman plots were constructed to compare echocardiographic and CMR data [21]. Results are presented as mean ^ 1 standard deviation. 3. Results Pre-operative LVMI ðn ¼ 39Þ, post-operative LVMI ðn ¼ 28Þ and both absolute reduction as well as percentage reduction of LVMI ðn ¼ 28Þ are shown in Table 1 for CMR and each of the echocardiographic formulae. Change in Fig. 1. Calculation of volumes using a CMR stack of contiguous short axis slices according to Simpson s method. ED end-diastolic images; ES end-systolic images. The epicardial borders are defined on the ED images, and the endocardial borders are defined on both the ED and ES images. This allows accurate measurement of the area of the blood pool and myocardium for each short axis slice, and subsequent calculation of ventricular volumes and mass. Note the signal loss (white arrow) around the sternum caused by sternal wires in this patient who had a sternotomy. This does not affect LV volume assessment. Fig. 2. Pre-operative (pre op) and post-operative (post op) LVMI values using CMR and each of the echocardiographic formulae. LVM by CMR was 43 ^ 26, 54 ^ 41 g/m 2 with the Bennett Evans (BE) formula, 42 ^ 32 g/m 2 with the ASE formula, 52 ^ 40 g/m 2 with the formula of Devereux with Penn correction (Penn) and 27 ^ 20 g/m 2 with the Teichholz formula. Individual changes are shown in Fig. 2. For all the scans performed both pre- and post-operatively, significant correlation was found between LVMI assessed with CMR and echocardiography using all the equations. However, discrepancy in a few high-end LVM measurements did influence the steepness of the slope of correlation to some extent (Fig. 3). Bland Altman plots showed that despite the significant correlation between LVMI using all four echocardiographic formulae and CMR and the effect of any particular points (Fig. 3), the Bland Altman limits were very wide for each formula, ranging from 99 to 152 g/m 2 (85 131%) (Table 2). There was no correlation between the change in LVMI after AVR assessed with any of the echocardiographic formulae and CMR, either when the absolute reduction in LVMI or the percentage reduction in LVMI was considered (Table 2 and Fig. 4). Bland Altman plots suggested that for increasing LVMI, BE, ASE and Penn echocardiography tended to overestimate reduction in LVMI compared to CMR, but overall the plots and wide Bland Altman limits of agreement confirmed the lack of any useful relationship between echocardiographic and CMR LVMI regression assessment (Fig. 4).

4 62 K. Rajappan et al. / European Journal of Cardio-thoracic Surgery 24 (2003) Table 1 Echocardiographic and CMR values of LVMI Pre-operative LVMI (N ¼ 39) Post-operative LVMI (N ¼ 28) Absolute reduction in LVMI Reduction in LVMI (%) CMR 134 ^ ^ ^ ^ 12 1D echocardiographic formulae BE 182 ^ ^ ^ ^ 23 ASE 147 ^ ^ ^ ^ 20 Penn 177 ^ ^ ^ ^ 21 Teichholz 112 ^ ^ ^ ^ 16 ASE, American Society of Echocardiography; BE, Bennett Evans; CMR, cardiovascular magnetic resonance; LVMI, left ventricular mass index. NB reduction in LVMI is not the difference between pre- and post-operative values, as only 28 patients underwent 1 year follow-up scans. Fig. 3. Scattergrams and Bland Altman plots of combined pre- and post-operative values of LVMI measured with CMR and echocardiography using the BE, ASE, Devereux with Penn correction (Penn) and Teichholz formulae. The line of unity (dashed line), linear regression line (solid line) and r values are shown on each scattergram.

5 K. Rajappan et al. / European Journal of Cardio-thoracic Surgery 24 (2003) Table 2 Differences, correlation and Bland Altman limits for mass and regression measurements CMR ECHO t-test P value Correlation Bland Altman limits Range of Bland Altman limits Absolute difference Difference (%) Absolute value Mean value (%) LVMI BE , to ASE , to Penn , to Teichholz 16 14, to Absolute reduction in LVMI post-avr BE to ASE to Penn to Teichholz to Absolute and percentage differences (CMR echocardiography) between the two techniques for LVMI and absolute reduction in LVMI after AVR, together with the Student s t-test and correlation coefficient. The Bland Altman limits of agreement are also illustrated (both absolute and as a percentage of the mean value). 4. Discussion The main results of this study are that, although there is some correlation between quantification of baseline LVMI with 1D M-mode echocardiography and CMR, the agreement is not acceptable in the assessment of LVM regression in individuals an end-point used widely in the literature while assessing the effect of therapeutic interventions such as AVR Assessment of regression of LVMI Although there was some correlation between echocardiography and CMR for LVMI assessment and agreement in mean differences in LVMI after AVR, there was no correlation in assessing regression in individuals. There are several key reasons for this. CMR has fundamental advantages over other imaging techniques, in that it produces accurate and reproducible 3D tomographic, static or cine images of high spatial and temporal resolution in any desired plane without exposure to contrast agents or ionising radiation. However, there is significant initial cost in setting up a CMR scanner and there are some people for whom it is not suitable e.g. those with permanent pacemakers. Compare this to M-mode echocardiography that, although widely available, cheap on a per scan basis, and portable, is operator and acoustic window dependent, and only measures a single segment assuming this is representative of the entire LV. The subsequently derived volume and mass measurements from formulae based on certain geometric assumptions do not hold true in the presence of regional variations. Studies on hypertensive patients with presumed concentric hypertrophy reported that only 8% of 165 hypertensive patients had typical concentric hypertrophy to which the various geometric echo formulae may be applied [22]. As the LV volume increases the LV becomes more spherical and the relation between length and diameter is altered. As a result, as the LV diameter increases, the 95% confidence interval of prediction of LV volume from the diameter rapidly increases [23]. 2D echocardiography overcomes some of these problems but still extrapolates data from limited views, which are very dependent on correct angulation of the probe, gain dependent edge identification and good endocardial border definition. Furthermore, there is significant intra-observer variability in measurement of wall thickness and internal diameter (6.9 and 2.3% for posterior wall thickness and end-diastolic diameter, respectively) [24]. These errors are cubed by the echo formulae and can result in changes in LV mass of approximately 8 15%. Errors are also exacerbated by the fast heart rate of some patients; a 2.7% decrease in end-diastolic diameter and a % increase in wall thickness occurs with each 10 beat/min increase in heart rate in normal subjects [25]. To date, most studies have used echocardiography to investigate regression of LVM after AVR. Although M-mode echocardiography is widely available and has been validated in the past [5], it is clear that newer 3D techniques such as CMR has overcome many of the inaccuracies associated with a 1D technique like M-mode. Previous studies have demonstrated that in comparison to CMR, echocardiography does not provide a reliable method of quantifying LV mass in individual patients [12]. In our study comparing a 3D with a 1D imaging modality, the lack of any relationship between the two techniques in the assessment of regression itself, suggests a particular limitation of M-mode echocardiography in assessing this marker of prognostic significance. M-mode echocardiography may allow semi-quantitative or qualitative assessment, but for accurate quantitative measurement CMR, and other 3D techniques such as 3D echocardiography are superior.

6 64 K. Rajappan et al. / European Journal of Cardio-thoracic Surgery 24 (2003) Fig. 4. Scattergrams and Bland Altman plots of absolute reduction in LVMI measured with CMR and echocardiography using the BE, ASE, Devereux with Penn correction (Penn) and Teichholz formulae. The line of unity (dashed line), and linear regression line (solid line) with equation, r and P values are shown on each scattergram. 5. Conclusions The implications of our findings are twofold. First, that in research studies the use of 3D techniques such as CMR should now be seen as the gold standard for LVM regression measurement. This may actually reduce costs because the increased accuracy allows the reduction of the number of subjects needed to detect significant changes. Secondly, future clinical studies and assessment should also involve the use of newer techniques for all the same reasons. References [1] Orsinelli DA, Aurigemma GP, Battista S, Krendel S, Gaasch WH. Left ventricular hypertrophy and mortality after aortic valve replacement for aortic stenosis. A high risk subgroup identified by preoperative relative wall thickness. J Am Coll Cardiol 1993;22: [2] Thomson HL, O Brien MF, Almeida AA, Tesar PJ, Davison MB, Burstow DJ. Haemodynamics and left ventricular mass regression: a comparison of the stentless, stented and mechanical aortic valve replacement. Eur J Cardiothorac Surg 1998;13:572 5.

7 K. Rajappan et al. / European Journal of Cardio-thoracic Surgery 24 (2003) [3] Hoffmann A, Burckhardt D. Patients at risk for cardiac death late after aortic valve replacement. Am Heart J 1990;120: [4] Jin XY, Pillai R, Westaby S. Medium-term determinants of left ventricular mass index after stentless aortic valve replacement. Ann Thorac Surg 1999;67: [5] Devereux RB, Reichek N. Echocardiograhic determination of left ventricular mass in man anatomic validation of the method. Circulation 1977;55: [6] Myerson SG, Bellenger NG, Pennell DJ. Assessment of left ventricular mass by cardiovascular magnetic resonance. Hypertension 2002;39: [7] Bottini PB, Carr AA, Prisant LM, Flickinger FW, Allison JD, Gottdiener JS. Magnetic resonance imaging compared to echocardiography to assess left ventricular mass in the hypertensive patient. Am J Hypertens 1995;8: [8] Semelka RC, Tomei E, Wagner S, Mayo J, Caputo G, O Sullivan M, Parmley WW, Chatterjee K, Wolfe C, Higgins CB. Interstudy reproducibility of dimensional and functional measurements between cine magnetic resonance studies in the morphologically abnormal left ventricle. Am Heart J 1990;119: [9] Bellenger NG, Davies LC, Francis JM, Coats AJS, Pennell DJ. Reduction in sample size for studies of remodeling in heart failure by the use of cardiovascular magnetic resonance. J Cardiovasc Magn Reson 2000;2: [10] Kaji S, Yang PC, Kerr AB, Tang WH, Meyer CH, Macovski A, Pauly JM, Nishimura DG, Hu BS. Rapid evaluation of left ventricular volume and mass without breath-holding using real-time interactive cardiac magnetic resonance imaging system. J Am Coll Cardiol 2001; 38: [11] Bellenger NG, Burgess MI, Ray SG, Lahiri A, Coats AJ, Cleland JG, Pennell DJ. Comparison of left ventricular ejection fraction and volumes in heart failure by echocardiography, radionuclide ventriculography and cardiovascular magnetic resonance; are they interchangeable? Eur Heart J 2000;21: [12] Grothues F, Smith GC, Monn JCC, Bellenger NG, Collins P, Klein HU, Pennell DJ. Comparison of interstudy reproducibility of cardiovascular magnetic resonance with two-dimensional echocardiography in normal subjects and in patients with heart failure or left ventricular hypertrophy. Am J Cardiol 2002;90: [13] Schmid FX, Keyser A, Djavidani B, Link J, Holmer S, Birnbaum DE. Left ventricular remodeling after pulmonary autograft aortic valve replacement: evaluation with color Doppler echocardiography and magnetic resonance imaging. Artif Organs 2002;26: [14] Gula G, Ahmed M, Thompson R, Radley-Smith R, Yacoub M. Combined homograft replacement of the aortic valve and aortic root with reimplantation of the coronary arteries (abstract). Circulation 1976;54:150. [15] Sahn DJ, DeMaria A, Kisslo J, Weyman A. Recommendations regarding quantitation in M-mode echocardiography: results of a survey of echocardiographic measurements. Circulation 1978;58: [16] Bennett DH, Evans DW. Correlation of left ventricular mass determined by echocardiography with vectorcardiographic and electrocardiographic voltage measurements. Br Heart J 1974;36: [17] Schiller NB, Shah PM, Crawford M, DeMaria A, Devereux R, Feigenbaum H, Gutgesell H, Reichek N, Sahn D, Schnittger I. Recommendations for quantitation of the left ventricle by twodimensional echocardiography. American Society of Echocardiography Committee on Standards, subcommittee on quantitation of two-dimensional echocardiograms. J Am Soc Echocardiogr 1989;2: [18] Teichholz LE, Kreulen T, Herman MV, Gorlin R. Problems in echocardiographic volume determinations: echocardiographic angiographic correlations in the presence of absence of asynergy. Am J Cardiol 1976;37:7 11. [19] Devereux RB, Lutas EM, Casale PN, Kligfield P, Eisenberg RR, Hammond IW, Miller DH, Reis G, Alderman MH, Laragh JH. Standardization of M-mode echocardiographic left ventricular anatomic measurements. J Am Coll Cardiol 1984;4: [20] Katz J, Milliken MC, Stray-Gundersen J, Buja LM, Parkey RW, Mitchell JH, Peshock RM. Estimation of human myocardial mass with MR imaging. Radiology 1988;169: [21] Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1(8476): [22] Ganau A, Devereux RB, Roman MJ, de Simone G, Pickering TG, Saba PS, Vargiu P, Simongini I, Laragh JH. Patterns of left ventricular hypertrophy and geometric remodeling in essential hypertension. J Am Coll Cardiol 1992;19: [23] Boudoulas H, Ruff PD, Fulkerson PK, Lewis RP. Relationship of angiographic and echographic dimensions in chronic left ventricular dilatation. Am Heart J 1983;106: [24] Gottdiener JS, Livengood SV, Meyer PS, Chase GA. Should echocardiography be performed to assess effects of antihypertensive therapy? Test retest reliability of echocardiography for measurement of left ventricular mass and function. J Am Coll Cardiol 1995;25: [25] Gosse P, Roudaut R, Dallocchio M. Is echocardiography an adequate method to evaluate left ventricular hypertrophy regression? Eur Heart J 1990;11: