MYOCARDIAL DEFORMATION IMAGING ON EXERCISE IN CHRONIC PRIMARY MITRAL REGURGITATION

Transcription

1 MYOCARDIAL DEFORMATION IMAGING ON EXERCISE IN CHRONIC PRIMARY MITRAL REGURGITATION A thesis submitted to the University of Manchester for the degree of Doctor of Medicine in the Faculty of Medical and Human Sciences 2012 RACHEL A. ARGYLE SCHOOL OF MEDICINE

2 CONTENTS LIST OF TABLES 7 LIST OF FIGURES 9 ABBREVIATIONS AND SYMBOLS 12 ABSTRACT OF THESIS 15 DECLARATION 16 COPYRIGHT STATEMENT 16 DEDICATION 17 ACKNOWLEDGMENTS 17 ABOUT THE AUTHOR 18 1 CHAPTER ONE - INTRODUCTION TO IMAGING TECHNIQUES AND THEIR APPLICATION INTRODUCTION Left ventricular contraction ADVANCED IMAGING TECHNIQUES USED IN THE ASSESSMENT OF LV FUNCTION Tissue Doppler imaging Speckle tracking echocardiography DEFORMATION IMAGING Strain and strain rate Twist (or torsion) Deformation imaging with stress CHAPTER 2 PRIMARY MITRAL REGURGITATION EPIDEMIOLOGY MITRAL VALVE ANATOMY 45 2

3 2.3 AETIOLOGY AND CLASSIFICATION OF MITRAL REGURGITATION MITRAL VALVE PROLAPSE Definition and incidence Pathology Clinical presentation and natural history ASSESSMENT OF THE SEVERITY OF MR Semi-quantitative measures Quantitative measures ADAPTIVE CHANGES OF THE LV TO SEVERE MR SEVERE MITRAL REGURGITATION WHEN TO INTERVENE? THE ROLE OF EXERCISE TESTING IN SEVERE MR Normal cardiovascular response to exercise Effects of exercise in patients with MR Uses of exercise testing in severe MR CHAPTER 3 - HYPOTHESIS AND AIMS OF THE STUDY 67 4 CHAPTER 4 - METHODS STUDY POPULATION STUDY DESIGN INVESTIGATIONS Resting echocardiography Exercise echocardiography Brain natriuretic peptide (BNP) Cardiopulmonary exercise testing ANALYSIS Echo analysis Statistical analysis CHAPTER 5 RESULTS OF BASIC MEASUREMENTS AT REST AND ON EXERCISE BASELINE CLINICAL CHARACTERISTICS 82 3

4 5.2 BASELINE ECHO CHARACTERISTICS At rest On exercise Comparison between rest and exercise within groups BASELINE CHARACTERISTICS ACCORDING TO CONTRACTILE RESERVE BASIC ECHO MEASUREMENTS ACCORDING TO CONTRACTILE RESERVE At rest and on exercise differences between groups Differences within groups from rest to exercise CHAPTER 6 BNP MEASUREMENTS AND CARDIOPULMONARY EXERCISE TESTING BNP MEASUREMENTS BNP in mitral regurgitation Results CARDIOPULMONARY EXERCISE TESTING Cardiopulmonary exercise testing in mitral regurgitation Results CHAPTER 7 LEFT VENTRICULAR TWIST ON EXERCISE MAGNITUDE OF TWIST AND TWIST RATE Between groups comparisons MR vs. controls Within groups comparisons: MR patients and controls rest vs. exercise Between groups comparisons CR subgroups vs. controls Within groups comparisons: contractile reserve subgroups and controls rest vs. exercise TIMING OF UNTWISTING Between groups comparisons MR vs. controls Within groups comparisons: MR patients and controls rest vs. exercise Between groups comparisons CR subgroups vs. controls

5 7.2.4 Within groups comparisons: contractile reserve subgroups and controls rest vs. exercise RATE OF UNTWISTING CORRELATIONS WITH STANDARD ECHOCARDIOGRAPHIC MEASURES REPRODUCIBILITY Reproducibility of ejection fraction measurements Reproducibility of twist measurements DISCUSSION OF RESULTS Main findings The importance of contractile reserve in mitral regurgitation LV twist and exercise The importance of untwisting The relationship between untwisting and contractile reserve CHAPTER 8 DEFORMATION ON EXERCISE PATIENT POPULATION MYOCARDIAL SAMPLING NOMENCLATURE OF RESULTS LONGITUDINAL DEFORMATION Between groups comparisons MR vs. controls Within groups comparisons MR patients and controls rest vs. exercise Between groups comparisons CR subgroups vs. controls Within groups comparisons rest vs. exercise in CR subgroups and controls RADIAL DEFORMATION Between groups comparison MR vs. controls Within groups comparisons MR patients and controls rest vs. exercise Between groups comparisons CR subgroups vs. controls Within groups comparisons rest vs. exercise in CR subgroups and controls

6 8.6 CORRELATIONS WITH STANDARD ECHOCARDIOGRAPHIC MEASUREMENTS Correlations at rest Correlations on exercise CORRELATIONS WITH TWIST VARIABLES DISCUSSION OF RESULTS Main findings Changes in deformation in chronic MR Deformation on exercise The relationship between strain and twist CHAPTER 9 CONCLUSION ASSESSMENT OF SUBCLINICAL LV DYSFUNCTION WITH DEFORMATION IMAGING STUDY LIMITATIONS WORK FOR THE FUTURE REFERENCES 154 Word Count: 41,975 6

7 LIST OF TABLES Table 1. Classification and causes of mitral regurgitation Table 2. Echocardiographic images used Table 3. Baseline clinical characteristics Table 4. Baseline echo characteristics at rest Table 5. Left ventricular volumes and ejection fraction on exercise Table 6. Change within groups in LV volumes from rest to exercise Table 7. Contractile reserve groups: baseline characteristics Table 8. Contractile reserve groups: standard echocardiographic measures on rest and exercise Table 9. Basic echo measurements - differences within groups from rest to exercise Table 10. Cardiopulmonary exercise test results Table 11. MR vs. controls on rest and exercise twist and twist rate Table 12. Rest vs exercise in MR patients and controls - twist and twist rate. 103 Table 13. CR subgroups vs. controls on rest and exercise twist and twist rate Table 14. Rest vs exercise in CR subgroups and controls - twist and twist rate Table 15. Timing of MVO and onset and peak of untwisting at rest and exercise: MR vs. controls Table 16. Timing of untwisting: rest vs. exercise in MR and controls Table 17. Timing of untwisting at rest and on exercise CR subgroups and controls Table 18. Timing of untwisting: rest vs. exercise in CR subgroups and controls

8 Table 19. Correlations between standard echocardiographic measures and twist variables at rest and on exercise in MR patients Table 20. MR vs. controls on rest and exercise - longitudinal deformation Table 21. Rest vs. exercise in MR patients and controls longitudinal deformation Table 22. Longitudinal deformation at rest and on exercise CR subgroups and controls Table 23. Longitudinal deformation at rest and on exercise CR subgroups and controls Table 24. Radial deformation at rest and exercise Table 25. Rest vs. exercise in MR patients and controls radial deformation 135 Table 26. Radial deformation at rest and on exercise CR subgroups and controls Table 27. Radial deformation at rest and on exercise CR subgroups and controls Table 28. Correlations between strain at rest and standard echocardiographic measures Table 29. Correlations between strain on exercise and standard echocardiographic measures Table 30. Correlations between strain and twist at rest and on exercise

9 LIST OF FIGURES Figure 1. Measurement of tissue velocities at multiple sites simultaneously using colour-tdi Figure 2. Unique pattern of speckles within each myocardial region Figure 3. Rotation traces obtained by speckle tracking of the base and apex of the LV Figure 4. Diagrammatic illustration of LV twist Figure 5. Normal twist curve Figure 6. Normal twist rate curve Figure 7. Schematic diagram of mitral valve anatomy Figure 8. Carpentier s mitral valve nomenclature Figure 9. Severe MR demonstrated by the area of the colour jet Figure 10. Measurement of MR flow rate using the PISA principle Figure 11. Tracing the short axis image Figure 12. Automatic generation of a tracking score Figure 13. BNP results by contractile reserve group Figure 14. Percentage predicted VO 2 max by contractile reserve group Figure 15. Averaged curves for twist at rest: MR vs. controls Figure 16. Averaged twist curves on exercise: MR vs. controls Figure 17. Averaged twist curves on rest and exercise MR patients Figure 18. Averaged twist rate curves on rest and exercise - MR patients Figure 19. Averaged twist curves on rest and exercise - controls Figure 20. Averaged twist rate curves on rest and exercise - controls Figure 21. Averaged curves for twist at rest CR subgroups and controls Figure 22. Averaged curves for twist on exercise CR subgroups and controls

10 Figure 23. Averaged curves for twist rate at rest: MR vs. controls Figure 24. Averaged curves for twist rate on exercise: MR vs. controls Figure 25. Averaged curves for twist rate at rest: CR subgroups and controls112 Figure 26. Averaged curves for twist rate on exercise: CR subgroups and controls Figure 27. Twist-volume loops at rest Figure 28. Twist-volume loops on exercise Figure 29. Correlation between PST on exercise and LA volume at rest Figure 30. Correlation between PST on exercise and regurgitant fraction Figure 31. Correlation between T-PST on exercise and LVEDV at rest Figure 32. Correlation between T-PST on exercise and LVESV on exercise. 120 Figure 33. Correlation between T-PST on exercise and change in EF Figure 34. Correlation between change in T-PUV and LA volume Figure 35. Between groups comparison of end-systolic strain Figure 36. Change in global longitudinal strain with exercise within groups Figure 37. Change in diastolic radial strain rate between rest and exercise Figure 38. Correlation between longitudinal SR S at rest and resting EF Figure 39. Correlation between longitudinal SR E and resting LVIDd Figure 40. Correlation between longitudinal SR E and regurgitant fraction Figure 41. Correlation between radial SR S on exercise and LVESV index at rest Figure 42. Correlation between radial SR E on exercise and regurgitant volume Figure 43. Correlation between longitudinal SR S at rest and PST at rest Figure 44. Correlation between longitudinal SR S at rest and change in PST from rest to exercise

11 Figure 45. Correlation between longitudinal SR E on exercise and time to peak systolic twist on exercise Figure 46. Correlation between radial SR E on exercise and peak systolic twist on exercise

12 ABBREVIATIONS AND SYMBOLS A Late (atrial) diastolic LV filling velocity A Atrial contraction tissue velocity AS AVC AVO BNP BSA BP BPM DSE DT E Ea ECG EDV EF EROA ESS ESV FPS FSV GLS HCM IVPG Aortic stenosis Aortic valve closure Aortic valve opening Brain natriuretic peptide Body surface area Blood pressure Beats per minute Dobutamine stress echo Deceleration time Early diastolic LV filling velocity Early diastolic tissue velocity Electrocardiogram End diastolic volume (LV) Ejection fraction Effective regurgitant orifice area End systolic strain End systolic volume (LV) Frames per second Forward stroke volume Global longitudinal strain Hypertrophic cardiomyopathy Intraventricular pressure gradient 12

13 IVSd LA LAD LV LVIDd LVIDs LVH LVOT MR MRI MVC MVO NYHA PA PISA PST PSTV PUV PW PWd PWs RCA RF ROI RV S Interventricular septal diameter at end diastole Left atrium Left anterior descending Left ventricle Left ventricular internal dimension at end diastole Left ventricular internal dimension at end systole Left ventricular hypertrophy Left ventricular outflow tract Mitral regurgitation Magnetic resonance imaging Mitral valve closure Mitral valve opening New York Heart Association Pulmonary artery Proximal isovolumic surface area Peak systolic twist Peak systolic twisting velocity Peak untwisting velocity Pulsed wave Posterior LV wall diameter at end diastole Posterior LV wall diameter at end systole Right coronary artery Regurgitant fraction Region of interest Regurgitant volume Systolic velocity 13

14 SD SR SRI STE SV TDI T-PST T-PUV VO 2 max V p VTI WMA ε έ ε N Standard deviation Strain rate Strain rate imaging Speckle tracking echocardiography Stroke volume Tissue Doppler Imaging Time to peak systolic twist Time to peak untwisting velocity Maximal oxygen uptake LV filling propagation velocity Velocity-time integral Wall motion assessment Lagrangian strain Strain rate Natural strain 14

15 ABSTRACT OF THESIS Thesis submitted to the University of Manchester by Rachel A. Argyle for the degree of Doctor of Medicine: Myocardial Deformation Imaging on Exercise in Chronic Primary Mitral Regurgitation May 2012 Background: Accurate assessment of left ventricular (LV) systolic function in chronic severe primary mitral regurgitation (MR) is important as the aim is to consider surgical repair prior to the onset of irreversible LV dysfunction. However this can be difficult to judge as conventional measures of LV function (such as ejection fraction, EF) may remain normal despite impaired LV contractility due to the increased preload of the condition. Advanced echocardiographic techniques, including deformation imaging, appear promising as they are less load dependent. As the earliest symptoms in severe MR are usually exertional, this study aimed to assess markers of LV deformation on exercise in patients with normal resting EF in order to try and identify the earliest signs of LV decompensation. Methods: Transthoracic echocardiography was carried out at rest and on submaximal supine exercise in asymptomatic patients with moderate to severe chronic primary MR and matched controls. Conventional contractile reserve (CR) as measured by EF change on exercise was used to subdivide patients into those with preserved (CR+) and abnormal (CR-) LV function. Myocardial strain and twist were assessed using the speckle tracking technique. Results: MR patients failed to show the normal enhancement in systolic twist on exercise. The onset and peak of untwisting were delayed in MR at rest and normalised on exercise in CR+ but not in CR-. Abnormalities in twist on exercise worsened with increasing resting preload. Longitudinal strain tended to increase normally on exercise in CR+ but not in CR-. Systolic longitudinal strain rate correlated with twist at rest and on exercise, whereas diastolic strain rate correlated with the timing of untwisting on exercise. Conclusion: Abnormalities in myocardial deformation are seen at rest and on exercise in patients with severe MR, particularly in those with decompensated LV function. This may contribute to the development of functional impairment with progressive disease. 15

16 DECLARATION No portion of the work referred to in the thesis has been submitted in support of an application for another degree or qualification of this or any other university or other institute of learning. COPYRIGHT STATEMENT i. The author of this thesis (including any appendices and/or schedules to this thesis) owns certain copyright or related rights in it (the Copyright ) and s/he has given The University of Manchester certain rights to use such Copyright, including for administrative purposes. ii. Copies of this thesis, either in full or in extracts and whether in hard or electronic copy, may be made only in accordance with the Copyright, Designs and Patents Act 1988 (as amended) and regulations issued under it or, where appropriate, in accordance with licensing agreements which the University has from time to time. This page must form part of any such copies made. iii. The ownership of certain copyright, patents, designs, trademarks and other intellectual property (the Intellectual Property ) and any reproductions of copyright works in the thesis, for example graphs and tables ( Reproductions ), which may be described in this thesis, may not be owned by the author and may be owned by third parties. Such Intellectual Property and Reproductions cannot and must not be made available for use without the prior written permission of the owner(s) of the relevant Intellectual Property and/or Reproductions. iv. Further information on the conditions under which disclosure, publication and commercialisation of this thesis, the Copyright and any Intellectual Property and/or Reproductions described in it may take place is available in the University IP Policy (see in any relevant Thesis restriction declarations deposited in the University Library, The 16

17 University Library s regulations (see and in The University s policy on Presentation of Theses. DEDICATION For my husband Chris, for his unfailing love and encouragement. ACKNOWLEDGMENTS I would like to gratefully thank the following: Dr Rhys Beynon, my colleague in the research department, for his support and help, as well as for his friendship and for making cups of tea, Dr Alex Borg, for his introduction to and tuition in the echo techniques involved, Mr Bruce Moreman, for development of the automated twist analysis program, Mr Keith Pearce and Dr Reza Aghamohammadzadeh for their help with the exercise echocardiograms, Mr Nigel Clayton, for his help with the cardiopulmonary exercise tests, The team in the Transplant Centre at the University Hospital of South Manchester, for analysis of the BNP samples, Dr Julie Morris, Head of Medical Statistics at the University of Manchester, for the power calculations, And finally, and importantly, Professor Simon Ray, my supervisor, for giving me the opportunity and resources to carry out this research, and for all his help with the project. 17

18 ABOUT THE AUTHOR I carried out this research at the University Hospital of South Manchester, during my Cardiology Specialist Registrar training in the North West Deanery. I had developed an interest in echocardiography and valvular heart disease during the earlier years of my cardiology training, and welcomed the opportunity to further my knowledge in these fields as well as to gain an understanding of the world of clinical research. This project was my first research experience. 18

19 1 CHAPTER ONE - INTRODUCTION TO IMAGING TECHNIQUES AND THEIR APPLICATION 1.1 Introduction The accurate assessment of myocardial function, particularly with respect to the left ventricle (LV), is necessary in many aspects of cardiology. In chronic valve disease this is especially important, as often decisions regarding the need for surgical intervention are based upon the finding of subtle degrees of LV dysfunction (1, 2). This is particularly the case in chronic primary mitral regurgitation (MR) where valve repair as soon as symptoms arise or signs of LV decompensation develop is currently thought to be the treatment of choice where feasible and appropriate (3-5). However, alternative schools of thought exist regarding surgical timing, which are discussed further in Chapter 2. The conventional measure of LV systolic function used in the clinical setting is the ejection fraction (EF). This is defined as the fraction of end-diastolic volume ejected from the heart with each contraction, and is calculated by the following formula: EF = End-diastolic volume (EDV) End-systolic volume (ESV) End diastolic volume EF has been shown to be a reasonably reproducible measurement (6-8), with prognostic implications in a variety of disease states including ischaemic heart disease and valvular heart disease (9-11). However, as a measure of LV function it has limitations. Firstly, quantitative echocardiographic assessment relies on geometrical models of LV shape, which can result in inaccuracies in measuring LV volumes and hence also in calculation of EF (12). Secondly, EF reflects LV systolic fibre shortening, and as such does not provide specific information on the other vectors of myocardial contraction (eg. circumferential contraction and twist) which are equally as important in the assessment of overall LV function. And finally, EF is affected by changes in preload and afterload (as well as heart rate), and therefore in valve disease in particular, an 19

20 EF within the normal range may not necessarily reflect normal myocardial function. There are generally three described components to myocardial systolic contraction; longitudinal, radial and circumferential contraction. It is recognised that in systole the apex of the heart remains relatively fixed, whilst the base moves downwards towards the apex: this is known as longitudinal function. At the same time, wall thickness increases ( radial contraction ), and rotation, or twisting, of the cardiac apex with respect to the base occurs. When viewed from the apex of the heart, rotation is seen to occur in an anticlockwise direction at the apex and a clockwise direction at the base, which overall gives rise to a net twisting motion of the LV as a whole (13). The reverse motions occur in diastole. Therefore, the challenge remains to be able to produce a simple, reliable and reproducible measure of LV function, preferably taking each of these various contractile components into account. Recent developments in echocardiography have given us new tools with which to do this, which can be applied both at rest and on exercise. Assessment of LV function on exercise is particularly useful and important as most cardiac conditions manifest symptoms only on exertion in the first instance. In the case of chronic primary MR in particular, the ultimate aim is to recognise the earliest signs of LV decompensation and to intervene before this becomes irreversible. If LV dysfunction on exertion can be identified at an early stage in the disease course then treatment may be able to be initiated before permanent damage occurs Left ventricular contraction LV contraction consists of a complex interplay of myocardial thickening, shortening and twisting in order to produce an efficient muscular pump. In the 1600s, Borelli likened the contraction of the heart to the wringing out of a rag (14), and this description has subsequently been corroborated by direct visualisation of the beating heart at surgery, and also by magnetic resonance imaging (MRI) (15). Currently, there remains controversy over the precise structure of myocardial fibres which enables the three components of contraction to occur. There 20

21 appear to be two main schools of thought, although it is recognised by both that muscle fibres align in differing orientations in different regions of the myocardium. Detailed dissection work by Greenbaum et al in the early 1980s showed that subendocardial fibres run longitudinally between LV apex and base whereas the mid-layer of the LV myocardium is comprised of circumferentially aligned fibres, predominantly at the base, with only a thin layer towards the apex (16). The subepicardial fibre arrangement is more complex, with a spiralling helical arrangement of fibres culminating at a vortex at the apex. The ongoing dispute comprises not so much this fibre orientation, but more the interrelationship between these layers ; is this a continuous band of muscle wrapping around the heart from pulmonary artery to aorta as championed by Torrent-Guasp and followers (17), or do the different myocardial fibres interweave in more of a mesh-like manner in order to function as a number of interrelated transmural units (18)? Both theories have their attractions and inconsistencies; Torrent-Guasp s myocardial band explains nicely a possible mechanism for an active (rather than passive) component to diastole, with sequential contraction of the band creating the effective untwisting of the ventricle and therefore generating the suction force for ventricular filling, but fails to consider the embryological basis to cardiac development or to correlate his proposed sequence of contraction of the muscle band to the anatomy of the electrophysiological pathways and the sequence of electrical activation of the LV (15). In contrast, Lunkenheimer s model emphasises the importance of connective tissue in helping to provide a dilating force required to counteract the muscular contraction of systole, but does not explain how this would produce the rapid untwisting and relaxation that is seen in early diastole. Regardless of the precise relationship between the myocardial fibres, it is recognised that the differing fibre orientations contribute varyingly to cardiac contraction and hence cardiac output. In particular, the longitudinal fibres, found predominantly in the sub-endocardium, play an extremely significant role in LV systolic contraction, despite the fact that they are relatively fewer in number than circumferentially arranged fibres. In normal hearts, during systole, LV short axis dimensions reduce by 25-40% despite the fact that the sarcomere shortens by only 10-12% (19). This suggests that shortening and transverse thickening of 21

22 the longitudinal fibres must contribute to radial as well as long axis function of the LV (20). In addition, longitudinal contraction plays an important role in ventricular filling with upwards motion of the atrioventricular ring reducing interatrial pressure and hence increasing atrial filling. It has also been shown that contraction of the longitudinal fibres occurs slightly earlier than circumferential fibres in normal subjects, beginning during isovolumic contraction (21). This normal sequence was altered in subjects with a variety of heart diseases, where longitudinal contraction was both delayed and reduced. It is postulated that the subendocardial location of the longitudinal myocardial fibres means that they are particularly prone to early dysfunction in a number of disease states due to easier susceptibility to ischaemic damage. Thus when assessing LV contractility, deterioration in longitudinal function may be an early marker of LV dysfunction. 1.2 Advanced imaging techniques used in the assessment of LV function Tissue Doppler imaging Tissue Doppler Imaging (TDI) is a relatively new tool enabling the measurement of regional or global tissue velocities. The Doppler technique is based upon the observations of Christian Doppler in 1842 that the observation of the frequency of light or sound waves depends upon the relative motion of the transmitter and receiver towards or away from each other. In echocardiography this principle was initially applied to the measurement of blood flow, where the change in frequency of the ultrasound wave as it is reflected back from moving blood (the frequency shift ) determines the direction and velocity of that blood flow. This can be represented on a display screen as either a spectral display showing a waveform of the direction and velocity of flow (spectral Doppler), or each velocity can be allocated a representative colour which can then be displayed on the screen as a colour map of the region being interrogated (colour Doppler). Initially when used in this context, filtering was applied so that only the high velocity, low amplitude frequencies from rapidly moving blood cells could be detected and signals from slower moving structures such as the myocardium 22

23 were filtered out (22). However, in the late 1980s it was recognised that this filtering process could be reversed, and by filtering out the low amplitude high frequency blood signals, the high amplitude low frequency signals from moving myocardium could be detected (23, 24). Doppler imaging, whether of blood flow or tissue, has certain limitations and these must be recognised. As the technique involves measuring frequency shifts towards or away from the ultrasound probe, only motion occurring in the parallel direction to the scanning angle is accurately measured. As the direction of motion becomes more tangential to the scan line (the angle of incidence ), measurement of frequency shift becomes increasingly underestimated, to the extent that once the angle of incidence exceeds about 20 degrees significant errors result, and once over 30 degrees information is very limited (25). Any motion perpendicular to the scan line is unable to be detected at all. This means that for practical purposes, when interrogating the myocardium from an apical window, only longitudinal movement can be measured. Scanning from a parasternal long or short axis approach enables measurement of radial tissue velocities in the posterior wall as this motion occurs within the line of interrogation of the probe. Analysis of radial velocities in the septum, which also lies perpendicular to the scanning line from parasternal windows, is complicated by the interaction between left- and right-ventricular components of this wall. TDI uses one of two different Doppler techniques pulsed wave or colour Doppler. Pulsed wave TDI (PW-TDI) measures peak instantaneous myocardial velocities at a specific site of interest ( the sample volume ), and is used on-line to obtain measurements whilst scanning. Temporal resolution is high (typically samples/s) (26), although in order to achieve this high frame rates of >150 frames/s should be used (27). Advantages of PW-TDI in addition to its high temporal resolution include the fact that tissue velocity information can be obtained instantly during a routine echocardiographic examination on most machines, but disadvantages are that the sample volume cannot be subsequently moved, and only one region of interest (ROI) can be interrogated at a time. Colour TDI is superimposed onto grey-scale 2-dimensional (2D) echo images, and analysis is carried out off-line in order to extract velocity data. The 23

24 myocardium is colour-coded either red or blue depending on its mean velocity and direction of movement. It is important to optimise 2D and colour gain settings and filters, and to maximise frame rates (preferably >100 frames/s) in order to obtain the best quality data. Advantages of this technique are that sample volumes can be repositioned, and velocities from several ROIs can be compared simultaneously, but the disadvantages of this are that postprocessing is required and the data cannot be obtained instantly. Also, the technique requires a degree of expertise and practice in order to optimise results. Figure 1. Measurement of tissue velocities at multiple sites simultaneously using colour-tdi Myocardial velocity data acquired from colour-tdi is generally 10-20% lower than that recorded from PW-TDI (28), and therefore for serial measurements the techniques are not interchangeable. It is also well recognised that myocardial velocities vary from region to region of the LV, with lateral wall velocities being higher than septal ones for both systole and diastole, and basal segments having a higher velocity than apical ones (29, 30). Myocardial 24

25 velocities also change with increasing age, with systolic velocity (S) and early diastolic velocity (E ) generally decreasing, and atrial contraction velocity (A ) increasing (29-31). One of the problems inherent in the interpretation of myocardial tissue velocities is that motion may be detected either as the result of active contraction of the region of the myocardium being examined, or as a consequence of passive movement due to a non-contractile segment being pulled by an adjacent contracting segment (32). This is known as tethering. Active contraction of the myocardium as a consequence of fibre shortening or lengthening is known as deformation. Using TDI to examine tissue velocities alone cannot differentiate between true deformation and tethering, but manipulation of colour-tdi data enables the derivation of deformation parameters such as strain and strain-rate Speckle tracking echocardiography 2-dimensional speckle tracking echocardiography (STE) is a relatively new technique which uses grey scale images to obtain data on myocardial deformation. The speckled pattern produced by acoustic backscatter from the ultrasound beam is tracked from frame to frame by automated software commercially available. This pattern of speckles is unique within each myocardial region (figure 2), and follows myocardial motion. Any change in the speckles represents myocardial deformation. The software extracts data on displacement, velocity, strain and strain-rate of each region from the change in speckle pattern. 25

26 Figure 2. Unique pattern of speckles within each myocardial region. In contrast to TDI, STE is angle independent as the speckles can be tracked in any direction (33). Therefore, from apical views deformation in both longitudinal and radial directions can be assessed, and in the short-axis view both radial and circumferential deformation can be measured. Circumferential deformation measurement allows the calculation of rotation of the ventricle, and hence also ventricular twist (34, 35). 26

27 Figure 3. Rotation traces obtained by speckle tracking of the base (top) and apex (bottom) of the LV. 2D-STE has been validated against sonomicrometry both in vitro and in vivo (36-38), with good correlation. It has also been assessed against tagged MRI (38, 39), with Cho s study showing moderate correlation between strain measured by MRI versus 2D-STE, although this was better than between MRI and TDI. Amundsen however, showed good correlation between MRI and 2D- STE-derived strain, with 80% of segments suitable for analysis. 27

28 2D-STE has also been compared with TDI in strain and velocity measurements in patients known or suspected to have ischaemic heart disease (39, 40). In both studies, results from the two techniques were found to be comparable. 3D-STE is a new application of the technique, which has recently been developed. This has potential advantages over 2D STE in that the speckled pattern of the myocardium can be tracked in three dimensions, reducing the likelihood of out-of-plane movement of the speckles. This should increase tracking accuracy and therefore produce more accurate measurements of deformation in all directions. The low frame rates traditionally associated with 3- dimensional echocardiography have limited the development of this application until very recently due to poor temporal and spatial resolution, but echocardiography systems that have overcome this are now commercially available. However, frame rates remain lower than those generally used for 2D- STE. A number of studies have validated 3D-STE measurements of LV volumes or strain against 2D-STE, MRI or sonomicrography (41-44). Seo et al measured strain in open-chested sheep with 3D-STE versus invasive sonomicrography and found good correlation between the techniques (44). 3D-STE has also been found to correlate better with MRI for the measurement of LV volumes than 2D- STE (42), and to have good test-retest reliability in this context (although less reliability for some strain measurements) (41). 3D-STE is quicker than 2D-STE when measuring strain, although the values obtained from each technique were significantly different in the study by Saito et al (43). As there is no clearly defined gold standard technique for the measurement of myocardial strain, it is difficult to ascertain from this whether 2D- or 3D-STE produce the most accurate measurements as the theoretical advantage of tracking speckles in three dimensions may be offset by the reduced temporospatial resolution resulting from the lower frame rates. Limitations of STE The primary limitation of STE relates to the need for excellent 2-dimensional image quality in order for the software to accurately track the speckle pattern and for adequate delineation of the endocardial border. Although smoothing of 28

29 the curves reduces extraneous noise, out of plane motion and reverberation artefacts can both interfere with the frame-by-frame analysis, affecting results (34, 45). The frame rates required for STE are lower than those needed for TDI, with optimal frame rates being somewhere in-between frames per second, depending on whether one or more walls are being examined (34, 45). Too low a frame rate results in under-sampling; this can also be a problem at high heart rates. Also if the frame rate is too low then the speckle pattern may have moved out of the area being searched by the tracking algorithm, which looks for the pattern in close proximity to where it was in the previous frame. Conversely, if the frame rate is too high then spatial resolution may be compromised, and tracking therefore impaired (46). 1.3 DEFORMATION IMAGING Strain and strain rate Concepts Strain is a relatively complex concept, which in echocardiography can be used as a measure of myocardial deformation. Lagrangian strain measures linear deformation of an object in relation to its original length, and can be defined as: ε= L-L 0 L 0 where ε is strain, L is the instantaneous length at the time of measurement and L 0 is the original length. With this definition, strain is a dimensionless ratio, and is usually expressed in percent. Applying this formula, it can be seen that stretching or lengthening of an object results in positive strain, and conversely shortening or compressing is negative strain. However it is also apparent that when an object has lengthened and then starts to shorten again, strain remains positive during this shortening as long as the instantaneous length is greater than the original length (and vice versa). This leads to the concept of natural, or Eulerian, strain, which relates strain to instantaneous length rather than original 29

30 length. It can be shown mathematically that natural strain (εn) is related to Lagrangian strain (ε) (47): ε= L L 0-1 and ε N =lnε+1 In terms of myocardial deformation, longitudinal contraction, or shortening, in systole can be well described by Lagrangian strain with L 0 being end-diastolic length. However, diastolic lengthening is more difficult to describe in this way due to L 0 being defined at the end of diastole (and therefore during diastole L 0 has not yet occurred). Diastolic deformation is more suited to being measured by strain-rate. Strain-rate (SR, έ) is the rate at which deformation occurs, and is equivalent to instantaneous strain (or change in strain) per unit time. The units used are /s (or s -1 ). SR is positive during lengthening and negative when shortening occurs, and therefore in diastole, when longitudinal fibres are lengthening, strain rate is positive. SR can be obtained by measuring the instantaneous velocities of two points, v 1 and v 2, and dividing the difference by the distance between them (the offset distance, d): ε= v 1-v 2 d However, in practice it is obtained by calculating the velocity gradient along the slope of the regression line of all velocities between the two points (48). TDI is able to provide information on tissue velocity, displacement, strain and strain-rate as the four parameters are all linked mathematically. The original velocity data obtained can be integrated with respect to time to obtain tissue displacement. If velocity data is spatially differentiated then strain-rate is obtained, and if strain rate is temporally integrated this produces strain. 30

31 temporal integration VELOCITY DISPLACEMENT spatial derivation spatial derivation STRAIN-RATE STRAIN temporal integration Speckle-tracking can also provide all of this data, although there are some potential differences to be aware of. When strain and strain-rate are obtained from TDI data, Eulerian or natural strain-rate is derived from the basic velocity data. STE, by contrast, measures Lagrangian strain directly. However, it has become conventional commercially when integrating natural strain-rate data from TDI to apply a correction so that Lagrangian strain is obtained. Therefore, in the same way, when deriving strain-rate from STE-acquired Lagrangian strain data, a similar correction is applied to obtain natural strain-rate. If this is not done then the results would not be comparable. Ultimately, this means that regardless of the method of data acquisition, strain is always presented as Lagrangian strain, and strain-rate is always presented as natural, or Eulerian, strain-rate (47). Limitations of the techniques The limitations of STE have already been discussed, and these apply equally to measurement of both tissue velocities and deformation. However, measurement of strain and strain-rate by TDI also has its problems which must be mentioned. Firstly, the technique is extremely prone to interference from noise, and this can impact significantly on the quality of the data. Care must be taken when acquiring the image in order to avoid areas of reverberation artefact, to avoid aliasing by increasing the colour scale (Nyquist limit), and to ensure the frame rate is as high as possible by optimising the imaging sector width and depth 31

32 (49). Secondly, as mentioned previously, any Doppler-based technique is only accurate when the scanning beam is aligned correctly with the region being interrogated (50). Thirdly, due to the high temporal resolution, spatial resolution is limited, and therefore tissue velocities may become contaminated with noisy signals from blood pool velocities. The sample volume therefore has to be tracked manually from frame to frame to ensure that it remains within the myocardium (51). Even with careful tracking, due to twisting of the ventricle the sample will almost inevitably move out of the scanning field at some point during the cardiac cycle, which becomes particularly important when measuring events in diastole (systole is less affected as peak strain occurs early in systole) (52). Respiratory motion and angle changes of the heart throughout the cardiac cycle can also affect the strain curves, so this also needs to be considered in acquisition. Strain and strain-rate imaging in practice Strain and SR measurements are relatively new developments in myocardial imaging. The derivation of SR from TDI was first described by Heimdal in 1998 (32), who demonstrated that the technique could be used to detect regional abnormalities of myocardial function. Both strain and strain-rate measures have subsequently been validated against other techniques as markers of cardiac contractility. Greenberg et al in 2002 examined TDI-derived systolic strain rate values against invasive measurements of peak systolic elastance (the goldstandard measure of LV contractility) in dogs in a variety of inotropic states (53). They compared strain-rate with peak systolic tissue velocity and found that although both measures altered in line with elastance according to changes in inotropic stimulation, strain-rate correlated significantly better with peak elastance than tissue velocity (r=0.94, p<0.01 vs. r=0.68, p<0.01). This is probably due to the elimination of artefact from translational movement of the myocardium. Edvardsen et al compared longitudinal TDI-strain and strain-rate with another non-invasive technique; 3-dimensional strain measured by tagged MRI (54). They compared healthy volunteers with patients both known and suspected ischaemic heart disease either at rest or during dobutamine stress echo (DSE). 32

33 Although the numbers of patients in some of the groups in this study were small, they concluded that strain and strain-rate derived from TDI correlated well with strain measured by MRI in a variety of settings within ischaemic heart disease, and confirmed Heimdal s previous work showing the ability to differentiate between variations in regional myocardial function which could potentially be clinically useful. More recently, Cho et al compared strain measured by both TDI and STE with MRI, although using a different MRI technique to Edvardsen (39). They assessed radial and circumferential strain in addition to longitudinal strain, and found STE to be more reliable and reproducible in these measures than TDI. However, overall strain by both TDI and STE correlated less well with MRI in their study than in Edvardsen s. Several reasons were proposed for this, including difficulties with echo image alignment compared with MRI, a time-lag of several days between the two imaging modalities being undertaken, and a higher proportion of their patients having regional wall abnormalities than Edvardsen s. In conclusion though, despite their correlations being more modest, they were still felt to be clinically acceptable. Studies have looked at the load-dependency of TDI-derived measures, including strain and SR. Urheim et al compared TDI-derived strain measurements with sonomicrometry in dogs during both ischaemia and volume loading (55). They found a good correlation between the two techniques, but noted that both myocardial velocities and strain were affected by increased preload. They also demonstrated the ability of strain in assessing regional myocardial function as opposed to myocardial velocities which were affected by ischaemia in distant segments and therefore less able to reflect regional function reliably. Several studies in haemodialysis patients have also demonstrated that tissue velocities are not completely load-independent (56-58). However Graham et al measured mitral annular velocities before and after haemodialysis and found no significant differences in their values despite other markers of diastolic function changing (59). Abali et al measured similar parameters in healthy volunteers before and after blood donation and also found no significant difference in mitral annular tissue velocities (60). Another study though also assessed healthy volunteers before and after preload-altering 33

34 manoeuvres and found that although mitral annular velocities were significantly altered by changes in preload, strain and velocities from within the LV myocardium itself were not (61). Wang et al also showed that myocardial strainrate measured during the isovolumic relaxation period was not affected by preload, although strain-rate during early diastolic filling was (62). As well as in ischaemic heart disease, there has also previously been much interest in the potential of strain rate imaging to detect subclinical disease in a variety of conditions. Several studies have shown that TDI and strain rate imaging can differentiate between pathological and physiological left ventricular hypertrophy (LVH) (63, 64), and one has also shown differences in systolic strain between athletes hearts and controls (65). Nagueh et al demonstrated its ability to predict the development of LVH in patients with the genetic mutation for hypertrophic cardiomyopathy who had not yet developed overt manifestations of the disease (66). Other conditions where abnormalities in tissue velocities or strain have been demonstrated include myotonic dystrophy (67), Becker s muscular dystrophy (68) and systemic sclerosis (69). However, what appeared initially to be a promising tool in the quest to detect early cardiac involvement or dysfunction in a wide range of situations has not translated into routine use in clinical practice. This is likely to reflect the difficulties inherent in the echo techniques involved, as well as perhaps limited clinical application regarding patients therapeutic management and outcomes Twist (or torsion) Twist events during the cardiac cycle Circumferential contraction of the LV gives rise to cardiac rotation. The alignment of the myocardial fibres mean that, when viewed from the apex, in systole the base of the heart rotates clockwise and the apex rotates anticlockwise, overall giving rise to a net twisting effect (13), Figure 4. The reverse untwisting occurs in diastole. By convention, anticlockwise rotation is defined as positive and clockwise as negative. LV torsion is calculated by subtracting rotation at the base from rotation at the apex, and as apical rotation is of higher magnitude than basal rotation, is normally positive overall in systole 34

35 and negative in diastole. Measurement of cardiac rotation or twist provides another dimension in the assessment of overall myocardial deformation. The terms twist and torsion are used interchangeably in the literature, although specifically, torsion refers to twist normalised for LV length. BASE SYSTOLE DIASTOLE APEX Figure 4. Diagrammatic illustration of LV twist. The normal pattern and timing of torsion has been well described by many authors (34, 35, 70-75). Kim et al in particular noted that in early systole, there is initially a short clockwise rotation at the apex and anticlockwise rotation at the base ( cocking ), before the predominant rapid twisting anticlockwise at the apex and clockwise at the base. This initial oppositely-directed rotation is thought to occur due to the earlier activation of subendocardial helical fibres compared with the epicardial ones (76). In normal subjects, peak systolic torsion (and thus the onset of untwisting) occurs on average just before aortic valve closure (AVC) (70, 77), Figure 5. LV systolic twist is physiologically important in the production of an adequate stroke volume (78), and in minimising the differences in end-systolic fibre stress, sarcomere shortening 35

36 and oxygen demand between endocardial and epicardial fibres within the LV wall (79-81). Figure 5. Normal twist curve. AVC: aortic valve closure, MVO, mitral valve opening There are then two distinct phases to untwisting (Figure 6); the first, rapid untwisting phase occurs during the isovolumic relaxation time prior to mitral valve opening, and around half of untwisting occurs during this time (70, 72, 77, 82). This early rapid untwisting is thought to be important in enhancing early ventricular filling by contributing to the development of an intraventricular pressure gradient (IVPG) between the base and apex of the LV (77), which is recognised to be a marker of early diastolic suction (83, 84). This improves LV filling even at high heart rates (72, 77, 82). It is also recognised that early ventricular untwisting is either delayed and/or reduced in a variety of cardiac pathologies, perhaps contributing to impaired cardiac relaxation and diastolic dysfunction (70, 71, 73, 74, 85-87). With the onset of LV filling, untwisting slows until equilibrium is again reached with the occurrence of the next systole. 36

37 Figure 6. Normal twist rate curve. Timing markers as per figure 4. Measurement of twist The presence of LV twist, although described by anatomists previously, was initially noticed invasively using cineradiography after radio-opaque markers had been surgically implanted into the myocardium (13). Tagged MRI studies followed (88), before early assessment of twist by echocardiography was carried out by Rothfeld et al in 1998 (89). This very simple and relatively crude measurement of twist involved measuring rotation of the insertion of the anterolateral papillary muscle around a point in the centre of the LV cavity in a mid-level short axis view. Results however, were comparable to the magnitude of twist found at a similar mid-ventricular level by either cine-radiography (90) or tagged-mri (88), and the same group went on to use the method again to assess torsion on exercise (91). Since then techniques have developed which enable measurement of twist by either TDI or STE. Using echo has advantages over MRI of accessibility and improved temporal resolution, and both echo modalities have been validated against tagged-mri and sonomicrometry, as well as against each other (34, 35, 92). The main problem with measuring rotation by TDI is its angle-dependency, 37

38 meaning that only those walls where rotation will occur within the line of interrogation of the ultrasound beam are able to be analysed. In practice, this means that only the septum and lateral walls can be assessed and therefore segmental rotation cannot be measured. However this method for assessing rotation, and therefore calculating twist, was employed by Notomi successfully in several of her studies including validation against MRI and on exercise (77, 82, 92, 93). STE has clear advantages over TDI in the measurement of twist, due to the lack of angle-dependency meaning that rotation can be measured in all walls in the short axis. The majority of recent studies assessing LV twist by echocardiography have therefore utilised this method (34, 35, 70, 75, 85, 86, ). The principle limitations relate to the previously mentioned need for good 2-dimensional image quality, leading to a feasibility on average of only 35-45% of subjects (75, 94). Clinical applications Many studies have assessed LV twist or torsion, in both normal subjects and in a variety of pathologies. Takeuchi et al took 118 normal healthy volunteers and obtained usable images in 113 of them using STE (98). They analysed the differing torsion patterns with increasing age, and noted that whilst peak torsion generally increased with age, untwisting was both delayed and had a reduced rate in older people, which would be consistent with the tendency towards diastolic dysfunction with increasing age. Notomi et al also assessed the effect of aging on torsion using TDI (93), and also found that, when normalised for LV length, LV untwisting reduced with increasing age. They noted that in infants, cardiac rotation was almost entirely unidirectional, with basal clockwise rotation developing throughout childhood and adolescence to a maximal wringing motion in middle age. In contrast however, Kim et al found no significant differences in torsion in older patients, but did note differences in basal rotation, with a reduction in both the amount of the initial anticlockwise rotation at the base and its duration (75). This correlated with alterations in the transmitral early diastolic filling wave velocity and deceleration time, and the authors therefore felt that this was a reflection of 38

39 decrease in diastolic function with increasing age. There was no such similar pattern noted with apical rotation, which was felt to be more related to systolic function. The correlation between torsion and LV systolic function has been examined in studies in subjects with dilated cardiomyopathy (85, 102). Tibayan et al measured torsion using cineradiography in sheep with dilated cardiomyopathy induced by rapid ventricular pacing and found that peak systolic torsion was significantly reduced compared with control conditions (the same sheep prior to the induction of cardiomyopathy) (102). They also noted that the early systolic clockwise cocking rotation was augmented in cardiomyopathy, and that peak twist was delayed until after AVC. Jin s study using STE in children with dilated cardiomyopathy compared with normal agematched controls had similar findings (85). They were able to distinguish that the reduction in twist arose primarily because of a significant reduction and disorganisation specifically in apical rotation. Aortic stenosis (AS) is associated with both systolic and diastolic dysfunction. Several studies have examined changes in torsion associated with AS (71, 73, 103). Sandstede et al examined patients with MRI-tagging both before and after aortic valve replacement (103). Despite the fact that prior to surgery patients had a slightly reduced EF compared to controls, overall systolic torsion was increased in patients with AS, primarily due to an increase in apical rotation. This was postulated to be a compensatory mechanism in order to maintain good systolic function in the presence of increased afterload, despite the reduction in radial contraction, and a reversal in both of these differences was seen 12-months post aortic valve replacement. Stuber and Nagel also both described this increase in systolic torsion in patients with aortic stenosis, and also both noted that untwisting in aortic stenosis is delayed, perhaps contributing to co-existing diastolic dysfunction (73). Diastolic changes in twist have been assessed in patients with LVH, as this is recognised to be associated with diastolic dysfunction. Takeuchi et al examined twist by STE in patients with LVH secondary to hypertension, and found that early diastolic untwisting was both delayed and reduced in these patients compared with a control group (86). These changes correlated with LV mass (r 2 =0.15, p<0.01). Systolic twist was unaffected. However, methodological 39

40 issues in their technique mean that these differences seen in early diastole may have been overestimated (94). In contrast to their findings, Rutz et al examined torsion by tagged-mri in patients with Fabry s disease (a condition associated with progressive LVH) and found that regardless of the presence or absence of LVH, peak systolic torsion was increased in the patients compared with the controls, and rapid early diastolic untwisting was faster, particularly in patients without LVH but also to an extent in those with LVH (104). They did not comment on the timing of this early untwisting. Other studies have also showed the early untwisting velocity to be unchanged in patients with diastolic dysfunction (82, 100), although in some cases its onset is delayed (82). Tibayan et al examined the change in LV torsion (by MRItagging) between acute and chronic MR by studying dogs 10 days and 3 months after MR had been induced surgically (87). They found that in the evolution from acute to chronic MR, peak systolic torsion was reduced and delayed (ie onset of untwisting was delayed) and early diastolic untwisting was reduced. Borg et al subsequently looked at patients with chronic primary mitral regurgitation and normal LV function with STE and found that both onset of untwisting and time to peak untwisting velocity were delayed in parallel with the degree of mitral regurgitation (70). Peak systolic torsion was not affected in their study; this difference may be due to the different progression of the chronic MR in their study population allowing a more physiological compensatory response of LV systolic function. Taking all these various findings overall, LV twist appears to play a key role in the transition between systole and diastole, with changes in both peak systolic torsion and early diastolic untwisting reflecting abnormalities in cardiac function in a number of cardiac pathologies. Of most interest is the fact that abnormalities in twist can be seen in patients without conventional echo manifestations of LV impairment, and clearly this could potentially become important in the monitoring and treatment of chronic conditions Deformation imaging with stress To be able to detect subclinical LV dysfunction is the holy grail of echocardiography in many situations, and this is undoubtedly the case in 40

41 chronic primary MR where decisions on the appropriate timing of surgery can be crucial. Imaging during exercise or pharmacological stress could help to identify the very earliest LV dysfunction which is undetectable at rest. In chronic MR specifically, there has already been work showing that an increase in EF on exercise of >4% predicts a good response to surgery (105), although EF itself is not the most sensitive marker of LV dysfunction, particularly in severe MR, due to its being affected by changes in loading conditions. Madaric et al found that even in those patients who did have an increase in EF of >4% on exercise, peak oxygen uptake and exercise capacity increased significantly after mitral valve repair surgery, suggesting that even asymptomatic patients with good LV contractile reserve did appear to have signs of LV dysfunction preoperatively which could not be detected (106). Changes in deformation imaging on stress could prove useful in view of the fact it is less load-dependent than EF. Lee et al found that peak-systolic and endsystolic strain at rest correlated with contractile reserve (by EF) on exercise (107), so logically the next step, particularly in view of Madaric s findings, would be to see if changes in strain and other deformation parameters on exercise could be an even more sensitive marker of LV dysfunction. Several studies have looked at strain imaging with exercise. In 2003, Davidavicius et al carried out a feasibility study assessing TDI-derived velocity and strain measurements in healthy volunteers undergoing either upright or supine bike exercise (10 in each group), and 10 patients undergoing dobutamine stress echocardiography (DSE) (108). They found that although the analysis was time-consuming, measurement of both regional strain and strainrate was feasible in patients undergoing DSE, but in both upright and supine bike exercise, image quality became significantly reduced and up to 40% of segments of the myocardium (particularly at the apex) could not be analysed. Nevertheless, they noticed in all three groups that whilst peak tissue velocities and strain-rate increased linearly with increasing heart rate, peak strain followed a biphasic pattern, initially increasing at low stress, but remaining constant or even decreasing slightly as heart-rate increased. This was felt to reflect the reduction in stroke volume at higher heart rates due to reduced LV filling, and was in agreement with earlier work by Voigt et al who suggested that SR rather 41

42 than strain itself was therefore more use in assessing myocardial function at stress (109). Reuss et al compared supine bicycle exercise with upright treadmill exercise in 18 healthy volunteers (110). They analysed TDI images from rest and peak exercise, and found no significant differences in strain at these stages (consistent with Davidavicius and Voigt), although peak tissue velocities and strain rate did both significantly increase (as found in the DSE studies). When adjusted for workload, there were no significant differences in the values obtained with upright versus supine exercise. Pierre-Justin et al also carried out a feasibility study using TDI-derived strain measurements obtained during supine bike exercise in healthy volunteers (111), and found similar results in terms of both patterns of change of velocities and strain / strain-rate on exercise and feasibility (72% of segments being analysable both at rest and peak exercise). This feasibility was further confirmed by Goebel et al in a similar, but slightly larger, study in 2007 (112), and overall these three studies provide a good database of normal reference values in young healthy people. Neilan et al assessed TDI-derived strain and STE-derived LV torsion in athletes before and after a short burst of high intensity exercise (113). They found that both peak systolic strain and systolic twist increased after exercise, although the pattern of diastolic filling changed with reduction in early diastolic filling and augmentation of late (atrial) filling. They assessed diastolic function using tissue velocities (as well as conventional echo measurements) and did not look at either strain/strain-rate or untwisting during diastole, which may have helped in the interpretation of the findings. Athletes were also examined in a study by Stefani et al comparing TDI-derived strain with STE-strain in basal segments of the ventricles (114). They assessed strain obtained by both echo techniques in basal segments of the left and right ventricles before and after hand-grip exercise in order to establish their concordance. Assessment of basal segments only was chosen in order to minimise the potential errors from the angle-dependence of TDI, and in this study strain measured by the two techniques was found to be both feasible and comparable. In line with the previous studies, no significant difference in strain between rest and exercise was found using either technique. 42

43 Few studies have assessed LV torsion on exercise. Notomi et al followed on from their previous work on torsion, and particularly diastolic untwisting, by looking at LV twist using TDI during supine bike exercise in normal subjects and patients with hypertrophic cardiomyopathy (HCM) and relating it to other markers of LV function, LV volumes, and IVPGs (82). They found that in normals both systolic and diastolic twisting were augmented more than LV longor short-axis function on exercise, and that peak untwisting preceded the development of the peak IVPG (thought to enhance early LV filling by a suction effect), which in turn occurred earlier than the peak early LV filling. Long- and short-axis functions were more temporally related to transmitral flow. In patients with HCM, untwisting occurred later both at rest and at exercise, and coincided with early LV filling rather than preceding it. Systolic twist, although greater at rest than in the normal controls, was not augmented as much on exercise, and untwisting velocities were lower both at rest and on exertion. These observations provide a link between systolic and diastolic events, and also provide a possible explanation for the diastolic dysfunction commonly seen in HCM. Burns et al have used STE to assess torsion on exercise, although feasibility was lower at only 42% (115). This was due to image degradation or an inadequate frame rate on exercise in a number of subjects, although in contrast to Notomi s study exercise was maximal rather than submaximal which may help to explain this relatively low feasibility. In agreement however, were their findings that both peak systolic torsion and peak twisting and untwisting velocities were increased with exercise. When results were analysed with respect to two age groups (mean ages 40 and 60 years), they found that at rest peak systolic torsion was higher in the older patients, although this failed to augment on exercise in the same manner as the younger group. Older patients also had less increase in twisting and untwisting velocities on exercise then the younger subjects. They did not assess the effect of age on the timing of twisting and untwisting events. In conclusion, data on deformation imaging, and particularly torsion, on exercise is limited. The majority of published studies in this area to date have examined normal subjects, but feasibility appears to be adequate enough to extend this 43

44 further to include cardiac disease. This has the potential to add further valuable methods in which to assess patients with a number of conditions, but in particular seems suited to the assessment of valvular disease which is often difficult using conventional imaging alone. 44

45 2 CHAPTER 2 PRIMARY MITRAL REGURGITATION 2.1 Epidemiology Valvular heart disease is common, with moderate to severe left-sided valve disease affecting an estimated 2.5% of the US population (116). Within this, mitral regurgitation (MR) has the highest prevalence, at an estimated 1.7% of the population overall, although varying with age from 0.1% in those aged up to 7.1% in people over 75 years. This burden of disease carries with it significant implications in terms of morbidity and mortality; the risk of sudden death with asymptomatic severe primary mitral regurgitation is estimated to be times normal, with an absolute risk of around 1-2.5% over 6 years (5). 2.2 Mitral valve anatomy The mitral valve is a complex structure, which for normal function relies on integrity of all of its individual components. These include the papillary muscles, the chordae tendinae, the valve annulus, and the leaflets themselves. Figure 7. Schematic diagram of mitral valve anatomy 45

46 Papillary muscles There are two papillary muscles; called anterolateral and posteromedial. Each of these may have several heads. They arise from the lateral and inferior walls of the left ventricle respectively. The anterolateral papillary muscle provides chordae to the anterolateral half of both mitral leaflets, and the posteromedial muscle similarly provides chordae to the posteromedial half of each leaflet. The posteromedial muscle is usually supplied by the right coronary artery, whereas the anterolateral papillary muscle usually receives its blood supply from both coronary arteries, which makes it less susceptible to ischaemic effects. Chordae Tendinae The chordae tendinae extend from the papillary muscles to the valve leaflets in a manner similar to the strings of a parachute. There are three levels of chordae; primary chordae attach to the leaflet edges, secondary chordae to the leaflet bodies, and tertiary chordae to the base of the posterior leaflet. Mitral annulus The mitral annulus is more complex than previously thought. It was initially thought to be a planar structure, but in actual fact has complex threedimensional geometry (117). It is fibrous; part of the fibrous skeleton of the heart which also includes the aortic and tricuspid valve annuli and the fibrous continuity between the anterior mitral valve leaflet and the aorta. Its shape can be likened to that of a Pringle crisp, as it is saddle-shaped, and the annulus can therefore appear either convex or concave towards the apex of the left ventricle depending on the angle from which it is viewed. Mitral valve leaflets The mitral valve has two leaflets, called anterior and posterior. The two points medially and laterally where the leaflets join are called the commissures; anterolateral and posteromedial respectively. When closed, the leaflets oppose each other centrally at the zone of coaptation. This coaptation zone is C- shaped, and the leaflets close against each other with an overlap of several millimetres, ensuring a good seal of the valve orifice when the valve is closed. 46

47 The anterior and posterior leaflets are structurally different. The anterior leaflet is larger and almost semi-circular in shape. The posterior leaflet, by contrast, is longer and shorter, with a greater length of attachment to the mitral valve annulus. The posterior leaflet has a scalloped appearance, which facilitates its division into three segments. Carpentier used this description in his widely used nomenclature, which calls the three posterior scallops P1, P2 and P3, with the adjacent anterior leaflet regions being named A1-3 respectively in a similar fashion, although they do not have the same morphological appearance. By convention, P1 and A1 lie laterally, and P3 and A3 medially (Figure 8). This enables accurate description of valve pathology to be communicated between sonographers, cardiologists and surgeons, and by ensuring that terminology is consistent, all parties are able to understand exactly which part of the valve may be defective. Figure 8. Carpentier s mitral valve nomenclature; LAA: left atrial appendage, Ao: aorta 47

48 2.3 Aetiology and classification of mitral regurgitation MR is the most common valvular abnormality. Its aetiologies are varied, but can broadly be divided into primary and secondary causes (see Table 1). Primary MR is that caused by either an abnormality of the valve leaflets themselves or of the subvalvular apparatus. Leaflet abnormalities can be either due to restriction and thickening of the valve, such as that caused by rheumatic valve disease, or alternatively due to redundant leaflet tissue, such as in mitral valve prolapse. Primary abnormalities of the subvalvular apparatus, for example papillary muscle rupture, are less common. Secondary, or functional, MR is due to distortion in the geometry or function of the mitral valve as a consequence of myocardial abnormalities. For example, a dilated cardiomyopathy may cause MR due to annular dilatation and failure of coaptation of the valve leaflets, and myocardial ischaemia can cause papillary muscle dysfunction, again leading to MR due to abnormalities in leaflet movement. It is recognised that functional MR may also be caused by dyssynchrony of the left ventricle, and in particular, dyssynchrony of the papillary muscles (118, 119). However, although in some patients with mitral valve prolapse fibrosis of the chordae and / or papillary muscles may be seen along with increased tissue velocity of the papillary muscles (120), no association with primary mitral regurgitation and dyssynchrony has been found (121). 48

49 Table 1. Classification and causes of mitral regurgitation Classification Example Pathology Primary MR Mitral valve prolapse Myxomatous degeneration Redundant valve tissue Ruptured chordate Rheumatic valve disease Leaflet thickening and restriction Commissural fusion Papillary muscle rupture Congenital abnormalities Myocardial infarction Cleft mitral valve Double orifice mitral valve Secondary MR Miscellaneous Annular dilatation Left ventricular wall motion abnormality Drugs Dilated cardiomyopathy Myocardial ischaemia / infarction 2.4 Mitral valve prolapse Definition and incidence Mitral valve prolapse (MVP) has been variably defined over many years, and hence its reported incidence has also varied depending on the definition in use at the time. Early studies relied on clinical findings of a non-ejection systolic click and mid or late systolic murmurs, phonocardiography, and later, M-mode echocardiography. A paper published in 1976 by Procacci et al examined 1169 healthy young women and found clinical signs of MVP in 6.3%, with echo findings to corroborate this in 81% of them (122). In the same year, Markiewicz et al examined 100 presumed healthy young women and diagnosed mitral valve prolapse in 10-18% of subjects based on the presence of phonocardiographic or echo abnormalities, or both (123). These are both clearly likely to be overestimates, and based on modern diagnostic criteria the incidence is thought to be around 2-3% of the population (124). 49

50 The current definition of MVP is based upon the work of Levine and colleagues in the 1980s, and their description of the saddle-shaped mitral annulus (117, 125, 126). The highest parts of the annulus are anterior and posterior, making the annulus concave upwards when viewed in the anteroposterior (AP) axis, and concave downwards when viewed medially-laterally. Therefore, in an apical 4-chamber echo view, which looks at the medial-lateral axis of the annulus, the leaflets can appear to bow back beyond the valve plane when closed when in actual fact they are normal. This means that diagnosis of MVP from a 4- chamber view lacks specificity and is no longer included in the standard definition. Echocardiographically, MVP is now defined as single or bileaflet prolapse of 2mm or more in the long-axis plane, with or without leaflet thickening (127). Having said that, some parts of the valve are not well seen from a long axis view and a full interrogation of all segments is needed if prolapse is suspected from the finding of significant mitral regurgitation Pathology MVP can occur with or without thickening of the valve leaflets, and the two presentations represent different pathological processes. In the 1960s, Barlow and colleagues first demonstrated that the syndrome of a mid-systolic click and murmur was associated with mitral regurgitation (128). However, it was Criley in 1966 who coined the term mitral valve prolapse and recognised that this was the cause of the regurgitation and not chord fibrosis as previously thought (129). Soon afterwards, the disease process was found to be degenerative rather than rheumatic as had been thought previously; the macroscopic appearance of thickened leaflets and elongated chordae was described, along with the histological findings of myxoid degeneration (130, 131). Sometime later the name Barlow syndrome was first used to describe this disease, particularly when associated with large billowing leaflets on echocardiography, and then in 1980 Carpentier differentiated it from the thinleafleted form of MVP not associated with excess tissue or leaflet billowing which he termed fibroelastic deficiency (132). Classic MVP, or Barlow syndrome, pathologically shows myxoid infiltration into the leaflets which results in disruption of the layers of the leaflets and collagen 50

51 alteration. The chordae are usually elongated, and can be thinned or, more commonly, thickened, and can rupture. The mitral valve annulus is frequently dilated. Most cases occur sporadically; although a familial (autosomal dominant) form is recognised. Multiple segments of the valve can be involved in the prolapse, and at surgery significant amounts of excess tissue are found often along with fissures at the base of the leaflets and calcification (133). The MR occurs as a consequence of the prolapsing edge of the leaflet segment, rather than the excess tissue itself, and can therefore have been there for many years prior to clinical presentation. By contrast, MVP due to fibroelastic deficiency is not associated with excess leaflet tissue, and instead is more commonly associated with chordal rupture with resultant severe MR. Connective tissue deficiency causes thinning of the valve leaflets and chordae, with secondary myxoid deposition being limited to the prolapsing segment alone. The cause is unknown, but is thought to possibly be age related as patients often present later in life and with a short history of symptoms or a murmur due to the ruptured chord. Most frequently only one valve segment is involved - often P2 - although any segment of either leaflet can be affected. Annular dilatation is less common than in Barlow s disease (133) Clinical presentation and natural history Clinical presentation is varied, often depending on the aetiology of the disease. Frequently an asymptomatic murmur leads to the first detection of the condition, or alternatively patients may complain of breathlessness, fatigue or palpitations. Many patients with MVP remain asymptomatic for years, and not all patients with the condition will progress to the stage of having severe MR. Complication and mortality rates for the MVP population overall are low, with both the Framingham study and more recent data suggesting adverse outcomes occurring no more frequently than in the general population (124, 134). However, once severe MR develops, the condition becomes less benign. Reported cardiac mortality rates in these patients varies from 36 ± 9% at five years (134) or around 6% per year (135) to 91 ± 3 % survival at 8 years (136). Complications include infective endocarditis, cerebrovascular events, atrial 51

52 fibrillation, heart failure, or sudden death, which are increased in patients with leaflet thickening, increasing age, and left ventricular impairment (127). In one study of 229 patients with flail mitral valve, mean rates of heart failure, atrial fibrillation and death or surgery at 10 years were 63 ± 8%, 30 ± 12%, and 90 ± 3% respectively (135). Incidence of sudden death in this patient group has been found to be around 1.8% per year, which is thought to be predominantly due to ventricular arrhythmias (137). These relatively high rates of adverse outcomes which develop as the disease progresses have led to the drive to investigate and attempt to clarify the optimal timing for intervention, as successful surgical repair of the valve has been demonstrated to restore normal life expectancy and improve symptoms (134, 138). 2.5 Assessment of the severity of MR The echocardiographic assessment of the severity of MR is fraught with difficulties, and many different parameters need to be taken into account before coming to a final conclusion as to the amount of regurgitation. In particular, MVP usually gives rise to highly eccentric jets of regurgitation which can prove even more difficult to assess. MR can be assessed semi-quantitatively or quantitatively, and a combination of each of these approaches is often needed Semi-quantitative measures Colour Doppler signal In general, the area of the regurgitant jet in the left atrium is directly proportional to the severity of the MR. This method of assessment was initially validated in several relatively small cohort studies with left ventriculography as the standard ( ). However, this is somewhat of a simplification, and there are a few additional considerations to take into account. Firstly, Colour Doppler is a map demonstrating the velocity of a regurgitant jet at a given moment rather than the absolute volume of regurgitation and is therefore not directly equivalent to the angiographic jet at ventriculography studies. Eccentric jets in particular, which 52

53 are constrained by the left atrial walls, may be underestimated if assessed purely on jet area by colour imaging as they do not spread into the LA in the same way as central or free jets. LA size itself can also make assessment of colour area more challenging. Indexing regurgitant jet area to left atrial area can potentially increase the accuracy of using this method to semi-quantify MR. Helmcke demonstrated that jet areas of < 20%, 20 40%, and >40% of left atrial area predicted mild, moderate, and severe angiographic regurgitation, respectively (139). These figures though need to be interpreted in the context of left atrial pressure and compliance, both of which can also affect jet area. Figure 9. Severe MR demonstrated by the area of the colour jet filling 74% of the left atrial area. Other issues to consider when interpreting colour Doppler regurgitant jet areas include instrument settings (eg. colour gain) and left ventricular pressure, ie. the 53

54 higher the LV pressure, the higher the velocity of the mitral regurgitant jet which will thus appear more severe on colour imaging. Continuous wave spectral Doppler The intensity of the spectral Doppler signal is proportional to the amount of red blood cells in the ultrasound beam. Therefore a large volume of regurgitation will produce a denser signal than a small volume jet. This can be used to aid assessment of the severity of MR (142). It is important to make sure that the ultrasound beam is well aligned with the direction of regurgitant flow to avoid underestimation of the signal. Mitral valve inflow velocity The forward flow velocity through the mitral valve, measured by pulsed-wave spectral Doppler at the mitral leaflet tips, can also help in determining the severity of MR (143, 144). This is related to the increased LA to LV gradient in MR due to the increased left atrial volume. A peak mitral valve E wave velocity >1.2m/s has a good sensitivity and specificity for detecting severe MR in the absence of mitral stenosis, particularly with normal LV ejection fraction. Pulmonary vein flow Pulmonary vein flow patterns, assessed by pulsed wave Doppler, can be helpful in the assessment of the severity of MR (145, 146). As left atrial pressure rises with increased regurgitant volume, systolic flow in the pulmonary veins reduces, and ultimately may completely reverse direction. These flow patterns are easier to see on transoesophageal rather than transthoracic echo, but often can be assessed to a degree on a standard transthoracic scan Quantitative measures Quantitative Doppler: regurgitant fraction and volume By measuring LV inflow at the mitral valve and LV outflow at the aortic valve, quantification of mitral regurgitant fraction and volume can be made (in the absence of aortic regurgitation). This technique has been described in detail by the American Society of Echocardiography (147). The area of the mitral annulus is calculated by measuring its diameter in the apical 4-chamber view and 54

55 applying πr 2. This assumes it is circular in shape, which is inaccurate, but the results using this assumption are satisfactory. This is then multiplied by the VTI of mitral inflow measured using PW Doppler at the level of the mitral annulus to give the forward stroke volume at the mitral valve. Similarly, forward stroke volume at the aortic valve is calculated from the LVOT area multiplied by the VTI in the LVOT. Mitral regurgitant volume is obtained by subtracting the outflow stroke volume from the inflow stroke volume, ie. Mitral regurgitant volume = (Area MV x VTI MV ) - (Area LVOT - VTI LVOT ) Regurgitant fraction expresses the regurgitant volume as a percentage of LV inflow, ie. Mitral regurgitant fraction = mitral regurgitant volume forward stroke volume at mitral valve There are a number of difficulties with this technique, the most obvious being the calculation of mitral valve and LVOT areas, as any error in measuring these diameters will be squared when turning a diameter into an area. Expressing the result as a regurgitant fraction rather than volume does at least have the advantage of cancelling out any error in the mitral annular area. The technique is also time-consuming, and requires a degree of expertise. However, it does remain sufficiently accurate for clinical use, and is considered the gold standard for Doppler quantification of valvular regurgitation. Effective regurgitant orifice area (EROA) Although the concept of effective regurgitant orifice area (EROA) is not a new one, having been described initially in 1952 by Gorlin and Dexter (148), the ability to calculate it non-invasively using echocardiography is a more recent development (149). Quantification of mitral regurgitation using this measure is attractive in that using the orifice area of a valve to assess its function is routine to all echocardiographers. However, measurement of EROA is not without its flaws. These include the fact that the regurgitant orifice may vary throughout the cardiac cycle for a number of reasons including loading conditions (150), LV size, and contractility (151). In addition, EROA has also been found to vary according to mitral valve pathology. Schwammenthal et al examined 56 patients 55

56 with mitral regurgitation of differing aetiologies and found clear patterns of temporal variation in MR flow rate and EROA (152). Whereas EROA may be relatively constant in rheumatic mitral valve disease, in functional MR due to LV dilatation or impairment more of a biphasic pattern was seen with peaks in early and late systole and a decrease in mid-systole. This was confirmed by other investigators (153), who felt that this was a reflection primarily of changes in transmitral pressure, but also to a lesser extent, changes in mitral annular area. In mitral valve prolapse, Schwammenthal found that most of the regurgitant flow occurred in mid to late-systole, with a corresponding marked increase in EROA during this time. The subgroup of patients with flail mitral valves however may have a more pansystolic duration of MR. Therefore using EROA alone to quantify MR may lead to over- or underestimation of severity of the lesion if the duration of regurgitant flow during systole is not taken into account. This is particularly the case in degenerative MR when although the peak EROA may be large, the duration of regurgitant flow may be relatively short and restricted to late systole. The EROA of the mitral valve can be calculated in one of two ways by echocardiography. Firstly, mitral regurgitant volume can be calculated by quantitative Doppler as described previously, and by dividing this by the VTI of the mitral regurgitant jet, an estimation of the mean EROA over a systolic cycle is obtained (147). However, this method is rarely used in clinical practice as discussed previously as it is relatively cumbersome and prone to measurement errors. Also, in view of the variability of EROA during systole, a measurement of the mean value may not be as clinically meaningful as its peak size. The alternative and more commonly used methodology is the flow convergence method, or Proximal Isovolumic Surface Area (PISA) method. This was first described in the early 1990s (149) and utilises the physics of flow acceleration towards a narrow orifice. As flow converges on the orifice from different directions, it accelerates up to its maximum velocity at the orifice itself. At a distance, r, from the orifice, the flow velocity will exceed the aliasing velocity of the colour Doppler scale, and will change colour abruptly. This will appear as an arc, or shell, of colour change on the 2D image, corresponding to a hemisphere of radius r in 3 dimensions. The area of this hemisphere (2πr 2 ) multiplied by the 56

57 velocity of flow at the point of the colour change of the shell (v) gives the instantaneous volume flow rate at the flow convergence zone. Figure 10. Measurement of MR flow rate using the PISA principle According to the principle of conservation of mass, this must equal the volume flow rate at the EROA, ie EROA x maximal flow velocity (Vmax - from CW Doppler through the MV). Therefore: EROA = 2πr2 v Vmax If this calculated EROA is multiplied by the VTI of the mitral regurgitant jet obtained with CW-Doppler, then mitral regurgitant volume is obtained. This technique is based on the assumption that the flow convergence zone is hemispherical in shape, which is clearly not always the case, so a correction can be made if necessary for cases of eccentric jets where the flow convergence zone is constrained by the wall of the LV and therefore not a full hemisphere. Despite this assumption, calculation of regurgitant volume in this 57