Halvorsen et al. multiple accelerations seen during one cardiac cycle. The acceleration signal can be integrated to obtain myocardial velocity, 1 bein

Transcription

1 British Journal of Anaesthesia 12 (1): (29) doi:1.193/bja/aen331 Advance Access publication November 19, 28 Detection of myocardial ischaemia by epicardial accelerometers in the pig P. S. Halvorsen 1 *, L. A. Fleischer 3, A. Espinoza 1, O. J. Elle 1, L. Hoff 3, H. Skulstad 2, T. Edvardsen 24 and E. Fosse 14 1 The Interventional Centre and 2 Department of Cardiology, Rikshospitalet University Hospital, N-27 Oslo, Norway. 3 Vestfold University College, Tønsberg, Norway. 4 The Faculty of Medicine, University of Oslo, Oslo, Norway *Corresponding author. per.steinar.halvorsen@rikshospitalet.no Background. We describe a novel technique for continuous real-time assessment of myocardial ischaemia using a three-axis accelerometer. Methods. In 14 anaesthetized open-chest pigs, two accelerometers were sutured on the left ventricle (LV) surface in the perfusion areas of the left anterior descending (LAD) and circumflex (CX) arteries. Acceleration was measured in the longitudinal, circumferential, and radial directions, and the corresponding epicardial velocities were calculated. Regional LV dysfunction was induced by for 6 s. Global LV function was altered by nitroprusside, epinephrine, esmolol, and fluid loading. Epicardial velocities were compared with strain by echocardiography during and with aortic flow and LV dp/dt max during interventions on global LV function. Results. induced ischaemia, shown by lengthening in systolic strain in the LV apical anterior region (P,.1) and concurrent changes in LAD accelerometer circumferential velocities during systole (P,.1) and during the isovolumic relaxation phase (P,.1). The changes in accelerometer circumferential velocities during were greater compared with the changes during the interventions on global function (P,.1). For the LAD accelerometer circumferential velocities, sensitivity was 94 1% and specificity was 92 94% in detecting ischaemia. Conclusions. Myocardial ischaemia can be detected with epicardial three-axis accelerometers. The accelerometer had the ability to distinguish ischaemia from interventions altering global myocardial function. This novel technique may be used for continuous real-time monitoring of myocardial ischaemia during and after cardiac surgery. Br J Anaesth 29; 12: Keywords: heart, ischaemia; measurement techniques, ultrasound; surgery, cardiovascular Accepted for publication: October 2, 28 Continuous monitoring of myocardial ischaemia during and after cardiac surgery remains a major problem. Electrocardiography (ECG) and haemodynamic measurements frequently fail to recognize myocardial ischaemia. 12 Transoesophageal echocardiography provides accurate assessments of ischaemia. 3 5 However, this technique requires a skilled operator, and only intermittent measurements are available. In the postoperative period, echocardiography monitoring is cumbersome and not routinely used. Hence, new methods for continuous real-time monitoring of myocardial ischaemia are needed. Systolic accelerations, assessed by accelerometers, have been shown to decline during coronary occlusion 6 8 and to correlate closely with ventricular dp/dt max during changes in inotropy. 9 Thus, accelerometers may enable continuous real-time monitoring of myocardial function during and after cardiac surgery. However, signals from the accelerations are difficult to interpret because of Declaration of interest. The three-axis accelerometer is patented by Rikshospitalet University Hospital, Oslo, Norway, for the use of detection of pre- and postoperative myocardial ischaemia and for monitoring of global myocardial function during and after cardiac surgery. The patent includes integration of the accelerometer into epicardial pacemaker wires. Regarding the patent, no relationships exist between Rikshospitalet University Hospital and any other companies. Engineer Ole Jakob Elle and Drs Erik Fosse and Per Steinar Halvorsen are patent holders. The other authors report no conflicts of interest. # The Board of Management and Trustees of the British Journal of Anaesthesia 28. All rights reserved. For Permissions, please journals.permissions@oxfordjournals.org

2 Halvorsen et al. multiple accelerations seen during one cardiac cycle. The acceleration signal can be integrated to obtain myocardial velocity, 1 being an established and sensitive parameter to measure myocardial ischaemia by ultrasound techniques. In this experimental study, changes in global myocardial function were induced by drugs and fluid challenges, and changes in regional myocardial function were made by temporary coronary artery occlusion. Changes in myocardial function were confirmed using speckle tracking echocardiography. The aim of the study was to validate an epicardial prototype three-axis accelerometer for the detection of regional myocardial ischaemia. We hypothesized that (i) during coronary occlusion, the accelerometer could identify changes in myocardial function also detected by speckled tracking echocardiography and (ii) the accelerometer could distinguish between global changes in myocardial function and regional changes induced by myocardial ischaemia. Methods The study was approved by the Rikshospitalet University Hospital Institutional Animal Care and Use Committee, and was carried out in accordance with Norwegian National Legislation on animal experimentation. Fourteen Norwegian landrace pigs of either sex, aged 3 4 months, with an average weight of 49.6 (range ) kg, were fasted overnight, but allowed free access to water. Before operation, they were sedated with i.m. ketamine 2 mg kg 21 and azaperone 3 mg kg 21 together with atropine.2 mg kg 21 to reduce ketamine induced salivation. Anaesthesia was induced with i.v. pentobarbital (2 3 mg kg 21 ) and boluses of morphine (.5 mg kg 21 ). Immediately, after induction of anaesthesia, a tracheotomy through a neck midline incision was performed, and the animals lungs were mechanically ventilated (Siemens KION 6., Solna, Sweden) with a mixture of room air containing 35% oxygen. Tidal volume and ventilation rate were adjusted to keep arterial PCO 2 close to 5.3 kpa. Anaesthesia was maintained with inspired isoflurane at a concentration of 1.% using a gas analyser (Siemens SC 9, Solna, Sweden) and i.v. morphine.15.2 mg kg 21 h 21, adjusted by guidance of the autonomic stress response of the pig. After a median sternotomy, the pericardium was split from the apex to base. An inflatable vascular occluder (In Vivo Metric, CA, USA) was placed around the proximal one-third of the LAD coronary artery. A prototype three-axis accelerometer (Kionix KXM52-15, Kionix Inc., USA) was sutured to the apical anterior wall of the left ventricle (LV) in the LAD perfusion area. A second accelerometer was sutured to the mid lateral wall in the perfusion area of the circumflex (CX) coronary artery (Fig. 1). The chest and the pericardium were left open, and the pig was placed in the supine position. Pressures and flow Under fluoroscopic guidance, a 5-Fr Millar micro manometer-tipped catheter (model MPC-5, Millar Instruments; Houston, TX, USA) was placed in the LV apical region through the right carotid artery. A second 5-Fr Millar catheter was positioned proximal to the aortic valve and a third in the left atrium. The pressure signal was zeroed against air and calibrated with a standard signal (1 mv¼1 mm Hg). The readings were compared with hydrostatic pressures for reference. A 5-Fr fluid-filled catheter was introduced through the right internal jugular vein to measure central venous pressure. All catheters were zeroed twice: before the interventions on global LV function and before measurements on regional LV function. Pressure and ECG data were processed through preamplifiers and digitized at 25 Hz for further analysis on a personal computer. An ultrasonic 16 mm flow-probe (Medistim, Oslo, Norway) was placed on the aorta for cardiac output measurements and a 4 mm probe (Medistim) was placed distally to the for LAD flow measurements. To ensure optimal signal quality, both probes were covered with gel during the interventions. Two-dimensional strain assessment using speckle tracking echocardiography Echocardiographic examinations were obtained by a Vivid 7 scanner (GE Vingmed AS, Horten, Norway). Conventional two-dimensional (2D) greyscale echocardiography was obtained from the LV short-axis at two LV levels, through the apical and basal regions. In addition, apical two and four chamber long-axis views were obtained. Longitudinal, circumferential, and radial LV function was assessed using 2D strain by speckle tracking echocardiography. This method calculates myocardial deformation from ultrasound speckles based on greyscale images, and has been validated as a reliable means of quantifying myocardial strain Mean frame rate was 62 (22) frames s 21. Strain curves of the LV anterior and lateral regions were assessed throughout the cardiac cycle and peak systolic strain was measured. The software program EchoPAC (GE Vingmed Ultrasound AS, Horten, Norway) was used for off-line analysis. We used a capacitive three-axis accelerometer (KXM52-14, Kionix Inc., Ithaca, NY, USA) mounted on a substrate and encapsulated in a silicon casing with outer dimensions of mm. 16 The accelerometer measurement range was +2 g and cross-sensitivity between axes 2% (Fig. 2). The accelerometer was calibrated using the earth s gravitational field. Pressures, acceleration, and ECG signals were recorded synchronously at sampling rates 25 5 Hz, using a NI USB 69 AD converter (National Instruments Inc., Austin, 3

3 Detection of myocardial ischaemia by accelerometers Basis LAD z y x control region K. Toverud y Apex z x apical region Fig 1 A schematic representation of the anterior view of the LV showing the position of two three-axis accelerometers, one in the LAD perfusion area and one in the CX perfusion area. The directions of the measured epicardial motions are indicated with arrows. X-arrow specifies longitudinal motion, Y-arrow circumferential motion, and Z-arrow radial epicardial motion. The level of is also indicated. TX, USA) and LabVIEW software (National Instruments Inc.). All signals were stored on a personal computer and the signals were analysed off-line with Matlab (The MathWorks Inc., Natick, MA, USA). Experimental protocol Global LV function was modified by infusing boluses of esmolol, nitroprusside, epinephrine, and colloid fluid. The interventions on global LV function were randomized by drawing lottery slips after the pig was anaesthetized. The medications and dosages were chosen based on their effects on preload, afterload, and contractility. Nitroprusside, esmolol, and epinephrine were selected because of their transient effect. Nitroprusside.1 mg was given to reduce preload and afterload. Preload was increased by infusion of 5 ml colloid fluid (Voluvene w ) within the range of duration of 5 min and 39 s and 8 min and 35 s. Epinephrine 1 mg and esmolol 1 mg were used to increase and reduce contractility, respectively. After the interventions on global LV function, regional myocardial ischaemia was obtained by left anterior descending (LAD) coronary artery occlusion for 6 s. LAD occlusion was always performed after the interventions on global LV function because of the risk of ventricular fibrillation during ischaemia and the possible modifying effects of ischaemia on the subsequent interventions. The occlusion time was based on pilot experiments, which showed marked changes in strain echocardiography after 6 s of. Prolonged occlusion time was associated with ventricular fibrillation. was verified using a LAD flow-probe (zero flow), cyanosis of the LAD supply region and by the occurrence of abnormal regional contraction (visual assessment and echocardiography). After each intervention on global function, the pig was allowed to recover for at least 15 min to return to haemodynamic baseline values. Occlusion of LAD was performed after a recovery period of at least 45 min. Data were acquired at all baselines and at the end of each intervention. To ensure metabolic stability during the experiment, blood samples for blood gas and haemoglobin (Hgb) analysis were extracted before the interventions on 31

4 Halvorsen et al. Baseline V sys V ivr velocity (cm s 1 ) V ivr V sys 3 Strain (%) 3 Pressure (mm Hg) 1 5 LV dp/dt (mm Hg s 1 ) ECG Fig 2 Representative curves of LV apical accelerometer circumferential velocity, echocardiography circumferential 2D strain and pressures (LV, aorta and left atrium), LV dp/dt max and ECG at baseline and after 1 min of. V sys, peak early systolic velocity; V ivr, velocity in the isovolumic relaxation phase. global LV function and before at the end of the experiment. Calculations Pressure-derived variables LV peak-systolic pressure, LV end-diastolic pressure, and LV positive and negative time derivates (LV dp/dt max and the LV dp/dt min ) were calculated. Systole was defined from the start of R on ECG to LV dp/dt min. Diastole was defined from LV dp/dt min to the start of R on ECG. The isovolumic relaxation phase (IVR) was defined from LV dp/dt min to the first diastolic LA and LV pressure crossover. The mean of three consecutive heart cycles was used for statistical analysis. Cardiac output (CO) and LAD flow were calculated as mean flow during over a time interval of 7 s. Myocardial strain In the longitudinal and circumferential direction, negative strain was defined as myocardial shortening and positive strain as lengthening. In the radial direction, positive strain was defined as thickening. Peak systolic strain was measured within the above defined systole and was calculated as a percentage of end-diastolic dimensions. The mean of three consecutive heart cycles was used for statistical analysis. Acceleration signals were high-pass filtered and integrated to velocity. 1 Peak early systolic ejection (V sys ) 11 was measured and V ivr was defined as the largest velocity spike during the IVR period. The accelerometer X-axis measured longitudinal-, Y-axis circumferential-, and Z-axis radial velocity (Fig. 1). Longitudinal velocity was defined positive from basis to apex, circumferential velocity was defined positive in the counter clockwise direction and radial velocity was defined positive towards the LV lumen. The mean of three consecutive heart cycles was used for statistical analysis. The accelerometer data were analysed without knowledge of reference method results. Data on intra- and interobserver (AE) variability were obtained by analysing a set of 2 randomly chosen accelerometer velocity curves during baseline and interventions. Statistical analysis The number of animals included was based on the results from pilot experiments. During ischaemia, a clinically significant change in accelerometer peak velocity was set to 5. cm s 21, with a standard deviation (SD) of 5. cm s 21. With a¼.5, this yielded power.9 with 11 pigs. Parametric statistical methods were used and data are presented as mean (SD). Repeated measurements ANOVA were used for all baseline values. No differences between baseline values were found for any measured variable. 32

5 Detection of myocardial ischaemia by accelerometers Therefore, for all interventions multiple Student t-tests with Bonferroni corrections of the P-values were used. Intra- and interobserver variation was analysed using Bland Altman method with 95% limits of agreement. 17 For the correlation analysis, Pearson correlation coefficient was calculated. Receiver-operating characteristic (ROC) curves were constructed to determine cut-off values for optimal sensitivity and specificity. Positive and negative predictive values were computed using a standard table analysis. A P,.5 was considered significant. Statistical analysis was performed using SPSS (Version 13, SPSS Inc., USA). Results Fourteen pigs underwent surgery. Two were excluded because of ventricular fibrillation. Thus, in the present study, data on 12 pigs are presented. Our experimental model remained stable during the experiment. No significant differences were observed in haemodynamic and accelerometer baseline values. In addition, ph, PCO 2, and lactate did not change significantly from start to end of the experiment. A small but significant change in Hgb was observed from the start [8.1 (1.2) g dl 21 ] to the end of the experiment [7. (1.7) g dl 21, P¼.31]. No significant ST-segment depression in the ECG lead II was observed (P¼.341) during. Only one animal showed ST-segment depression..1 mv. Haemodynamic variables During significant changes were observed in all haemodynamic variables, except for heart rate (Table 1). LV peak-systolic pressure, LV dp/dt max and aortic flow decreased, whereas LV end-diastolic pressure increased, indicating LV failure. Myocardial strain Regional strain measurements were obtained in 1 of 12 animals. As a result of logistic reasons, echocardiography was not performed in two animals. In the anterior LV region, significant and marked changes in strain were observed in all directions during (Table 2). In the circumferential and longitudinal directions, strain showed lengthening, indicating severe myocardial Table 1 Haemodynamic variables during interventions on regional and global LV function (n¼12). LV, left ventricle; LV dp/dt max, positive time derivative of LV pressure; LAD, left anterior descending artery. Values are mean (SD). P-values are Bonferroni corrected. *P,.5 and **P,.1 from baseline Intervention baseline effect Magnitude of change Heart rate (beats s 21 ) 87 (9) 88 (9) 21 (2) LV peak systolic pressure (mm Hg) 98 (13) 84 (12)** 13 (6) LV dp/dt max (mm Hg s 21 ) 1599 (329) 1254 (22)** 345 (226) LV end-diastolic pressure (mm Hg) 15 (4) 2 (6)** 24 (3) Aorta flow (litre min 21 ) 4.9 (2.) 4.4 (2.)**.5 (.4) LAD flow (ml min 21 ) 25 (12) ()** 25 (12) Epinephrine Heart rate (beats s 21 ) 83 (11) 133 (32)** 25 (25) LV peak systolic pressure (mm Hg) 95 (16) 145 (15)** 249 (21) LV dp/dt max (mm Hg s 21 ) 1552 (336) 4372 (2325)* 2282 (2383) LV end-diastolic pressure (mm Hg) 15 (5) 14 (6) 1 (5) Aorta flow (litre min 21 ) 4.3 (1.7) 6.8 (2.9)** 22.5 (1.4) LAD flow (ml min 21 ) 24 (14) 6 (44) 236 (32) Esmolol Heart rate (beats s 21 ) 83 (13) 8 (1)* 4 (4) LV peak systolic pressure (mm Hg) 93 (12) 75 (8)** 19 (6) LV dp/dt max (mm Hg s 21 ) 1547 (44) 911 (222)** 636 (248) LV end-diastolic pressure (mm Hg) 13 (3) 14 (3) 21 (1) Aorta flow (litre min 21 ) 4.5 (2.1) 3.9 (2.1)**.6 (.3) LAD flow (ml min 21 ) 21 (13) 16 (1)** 5 (4) Fluid loading Heart rate (beats s 21 ) 85 (1) 85 (1) 1 (4) LV peak systolic pressure (mm Hg) 92 (13) 1 (12)** 28 (4) LV dp/dt max (mm Hg s 21 ) 152 (39) 166 (235) 286 (143) LV end-diastolic pressure (mm Hg) 12 (3) 24 (7)** 211 (5) Aorta flow (litre min 21 ) 4.1 (1.5) 5. (1.6)** 2.9 (.4) LAD flow (ml min 21 ) 21 (1) 31 (1)** 21 (7) Nitroprusside Heart rate (beats s 21 ) 82 (12) 85 (12)* 23 (3) LV peak systolic pressure (mm Hg) 94 (14) 78 (15)** 16 (5) LV dp/dt max (mm Hg s 21 ) 1536 (336) 1312 (381)** 224 (13) LV end-diastolic pressure (mm Hg) 14 (3) 9 (2)** 5 (3) Aorta flow (litre min 21 ) 4.4 (2.) 4.4 (2.5). (.8) LAD flow (ml min 21 ) 24 (13) 22 (13)* 2 (2) 33

6 Halvorsen et al. Table 2 Left ventricle 2D strain during (n¼1). LV, left ventricle. Values are mean (SD). P-values are Bonferroni corrected. *P,.5 and **P,.1 from baseline LV region 2D strain (%) Absolute value baseline Magnitude of change Apical Circumferential (4.4) 12.2 (11.)** (7.4) anterior Longitudinal 22. (4.6) 7.6 (4.)** (5.3) Radial 64.9 (12.2) 14.1 (23.1)** 5.8 (23.3) Lateral Circumferential (5.9) (6.4).3 (2.1) Longitudinal (3.1) 22.3 (5.4).6 (3.8) Radial 55.8 (8.5) 56.7 (12.2) 2.9 (11.1) Echocardiography circumferential 2D strain (%) = Baseline = ischaemia with paradox movement of the anterior LV region. In the control region, no significant changes in strain were observed (Table 2). LAD accelerometer velocities In Figure 2, showing LAD accelerometer circumferential velocity curves, a decrease in circumferential V sys and an increase in V ivr during can be seen. Circumferential V sys decreased from 14.1 (3.4) to 6. (2.5) cm s 21 (P,.1) during, whereas in the longitudinal and radial directions V sys did not change significantly (Table 3). In the LV anterior region, similar changes in circumferential strain and circumferential V sys were observed in all animals (Fig. 3). During ischaemia, a marked change in circumferential V ivr was observed [25.3 (3.3) to 7.7 (6.6) cm s 21, P,.1] (Table 3). V ivr changed from a negative to a positive value in all animals during. Similar trends in V ivr were observed in the longitudinal and radial directions, but these changes were not significant. Thus, the dominant change during ischaemia in myocardial LAD accelerometer circumferential peak systolic velocity (cm s 1 ) Fig 3 Individual changes in circumferential 2D strain and accelerometer circumferential peak systolic velocity for the LV apical anterior region, at baseline and after 1 min of. function assessed by the accelerometer was exhibited in circumferential V ivr. CX accelerometer velocities During LAD, occlusion V sys was significantly reduced in all directions (Table 4). For V ivr a significant increase was observed in the longitudinal (P,.1) and radial axes (P¼.5), but not in the circumferential direction. Effects of interventions on global cardiac function Haemodynamic variables Significant and typical haemodynamic changes were seen during all interventions on global cardiac function (Table 1). LV end-diastolic pressure increased significantly Table 3 Velocity signals from the LAD accelerometer during interventions on regional and global LV function (n¼12). IVR, isovolumic relaxation period; LAD, left anterior descending artery. Values are mean (SD). P-values are Bonferroni corrected. *P,.5 and **P,.1 from baseline Axis Interventions Peak systole (cm s 21 ) IVR (cm s 21 ) baseline effect Magnitude of change baseline effect Magnitude of change Circumferential Longitudinal Radial LAD 14.1 (3.4) 6. (2.5)** 8.2 (2.) 25.3 (3.3) 7.7 (6.6)** (6.8) occlusion Epinephrine 13.7 (2.9) 25.2 (8.1)** (7.4) 25.6 (3.4) 25.8 (4.3).2 (3.) Esmolol 14. (3.1) 1.2 (3.3)** 3.7 (1.) 24.6 (3.5) 21.8 (2.7)* 22.3 (3.3) Fluid loading 12.8 (3.3) 15.9 (3.5)** 23.1 (1.6) 24.4 (3.5) 27.9 (3.5)** 3. (2.1) Nitroprusside 14. (3.3) 15.4 (3.6) 21.5 (2.) 26. (4.1) 26.5 (5.1) 1.4 (2.6) LAD 12.4 (3.6) 1.4 (5.) 2. (4.1) 27.2 (4.4) 25. (4.1) 22.2 (6.6) occlusion Epinephrine 12. (3.4) 21.8 (7.)** 29.8 (6.8) 27.5 (4.2) 21.7 (5.3) 3.2 (5.8) Esmolol 13.5 (3.2) 9.6 (3.2)** 3.9 (1.2) 26.8 (3.5) 24.6 (3.9) 22.3 (3.1) Fluid loading 13. (3.5) 13. (4.8). (3.) 26.8 (3.9) 27.3 (4.1).5 (3.9) Nitroprusside 12.2 (3.5) 13.8 (3.2) 21.5 (2.1) 26.7 (3.7) 26.2 (4.1) 2.5 (3.4) LAD 3.7 (1.9) 1.9 (2.3) 1.8 (2.6) 21.2 (6.4).5 (5.) 21.8 (3.) occlusion Epinephrine 3.6 (2.5) 8.6 (2.9)** 25. (4.3) 21.2 (7.) 22.3 (6.9) 1.1 (5.) Esmolol 3.6 (2.4) 2.9 (1.7).7 (.9) 21.2 (6.2) 21.3 (3.8).1 (3.4) Fluid loading 3.2 (2.4) 4.1 (2.2) 2.9 (1.1) 2.8 (6.2) 22.2 (6.9)* 1.4 (1.4) Nitroprusside 3.7 (2.5) 3.5 (2.5).2 (1.4) 2.4 (6.3).6 (6.4) 21. (1.4) 34

7 Detection of myocardial ischaemia by accelerometers Table 4 Velocity signals from the CX accelerometer during. IVR, isovolumic relaxation period. Values are mean (SD). P-values are Bonferroni corrected. *P,.5 and **P,.1 from baseline CX accelerometer Axis Circumferential Longitudinal Radial Velocity (cm s 21 ) Absolute value baseline Absolute value LAD occlusion Magnitude of change Peak 5.6 (2.) 3.2 (2.6)* 2.4 (2.) systole IVR 21.1 (2.7) 2. (5.1) 23.1 (4.8) Peak 8.6 (2.7) 5.8 (2.6)** 2.9 (2.2) systole IVR 2.7 (2.8) 3.1 (4.)** 23.8 (2.2) Peak 3.9 (2.1).7 (2.8)* 3.2 (2.6) systole IVR 2.3 (2.5) 2.1 (2.3)* 22.2 (2.4) during fluid loading but, in contrast to the increase in LV end-diastolic pressure during, fluid loading was associated with an increase in LAD flow, LV pressure, and aortic flow, and did not indicate LV failure. velocities V sys in the circumferential direction was the only measure that was able to reflect global cardiac function in a wide range of different interventions (Table 3). During epinephrine infusion, systolic velocities increased substantially in all directions. With esmolol infusion, a decrease in V sys was found in the longitudinal and circumferential directions, while V sys in the radial direction remained unchanged. During volume loading circumferential, V sys increased significantly, whereas longitudinal and radial velocity did not demonstrate significant changes. V sys did not change significantly during nitroprusside infusion. Similar to, esmolol induced a decrease in accelerometer circumferential V sys and an increase in V ivr. However, despite the greater haemodynamic changes induced by esmolol, the magnitude of changes in accelerometer velocities was much less compared with changes caused by (Table 1). In particular, this was evident for V ivr. Volume loading and ephinephrine infusion induced opposite changes in circumferential velocities, while nitroprusside caused no significant changes. The effects on LAD accelerometer circumferential velocities induced by interventions affecting regional and global function were compared using repeated measurements ANOVA with Bonferroni correction of P-values. The absolute value of accelerometer V sys during ischaemia was significantly different from all corresponding absolute values obtained from interventions on global myocardial function; nitroprusside (P,.1), epinephrine (P,.1), esmolol (P¼.2), and fluid loading (P,.1). The absolute value of V ivr during also differed significantly from absolute V ivr values obtained from the interventions with nitroprusside (P,.1), epinephrine (P,.1), esmolol (P¼.4), and fluid loading (P,.1). For the interventions on global function, changes in circumferential V sys correlated strongly with changes in CO (r¼.81, P,.1) and with changes in LV dp/dt max (r¼.73, P,.1). Results for the CX accelerometer were similar to those found for the LAD accelerometer (data not presented). Sensitivity and specificity ROC analysis was performed on all baseline- and interventional-values on LAD accelerometer circumferential V sys (n¼12). Ischaemia was detected with a sensitivity of 94% and a specificity of 92% using a cut-off value of 8.7 cm s 21. Applying this cut-off value for circumferential V sys, the positive predictive value was 67% and negative predictive value was 1%. ROC analysis on circumferential V ivr demonstrated sensitivity of 1% and specificity of 94% (cut-off value of.3 cm s 21 ) for detecting ischaemia. Using.3 cm s 21 as a cut-off value for circumferential V ivr, positive predictive value was 63% and negative predictive value was 1%. Inter- and intraobserver variability Intraobserver variability [mean difference (95% confidence interval) (2 SD)] for the circumferential V sys and V ivr was.2 (2.4.9) (1.8) cm s 21 (r¼.94) and.5 ( ) (3.2) cm s 21 (r¼.98), respectively. Interobserver variability for the circumferential V sys and V ivr was.1 (2.1.3) (.6) cm s 21 (r¼.99) and 2.2 (21..5) (2.1) cm s 21 (r¼.99), respectively. Discussion The present study demonstrates the potential of an epicardial three-axis accelerometer in detecting myocardial ischaemia. For the accelerometer, a marked reduction in circumferential systolic velocity together with a negative to positive shift in velocity during IVR was characteristic for regional ischaemia and was not found during interventions changing global LV function. Our study also demonstrated that the accelerometer circumferential systolic velocity was a marker of global LV function. Compared with echocardiograpic methods, the main advantage for the accelerometer is that it may enable operator independent continuous real-time monitoring of myocardial ischaemia. In particular, in the early postoperative phase, there is a need for continuous and sensitive methods, since echocardiography is not routinely performed in this period. As a potential method for continuous perioperative monitoring of ischaemia, the important first step was to investigate whether accelerometers were able to detect ischaemia with acceptable accuracy. Sensitivity of the accelerometer in detecting ischaemia Reduced myocardial function during coronary occlusion was confirmed by our strain measurements. Paradoxical 35

8 Halvorsen et al. movement of the LV anterior region during systole is a sign of severe myocardial ischaemia The reduction in myocardial function during coronary occlusion was also detected by the accelerometer. Our study clearly demonstrated that the accelerometer s circumferential systolic and IVR velocities were sensitive markers of myocardial ischaemia. The technique provided velocity traces of high quality, with little inter and intraobserver variation, and was superior to ECG ST-segment analysis that is at present the only continuous method to detect ischaemia. Our system enabled real-time presentation of epicardial velocity, thereby enabling the possible use of automated algorithms for analysis and interpretation of signals to detect ischaemia. Thus, in this study, the principle of using accelerometers for continuous real-time monitoring of myocardial ischaemia was documented. Our observations of altered epicardial velocities by accelerometry during ischaemia are in accordance with previous reports using the ultrasonic technique. Because of the fundamental differences between the two techniques in measuring myocardial function, this relationship was not obvious. Echocardiography tissue velocities are assessed within the myocardium, while the accelerometer measures epicardial heart wall motion. For the accelerometer in the LV apical region, the different epicardial velocities were affected unequally during and significant changes were only observed in the circumferential direction. A reason for this discrepancy between the accelerometer velocities during could be the use of an open chest model. In this model, reflecting the clinical setting of cardiac surgery, the pericardium was not sutured. Loss of pericardial constraint causes abnormal longitudinal apex motion, 2 21 whereas LV rotation remains unaffected. 15 In the apical region, myocardial contraction is dominated by circumferentially orientated fibres and therefore it was not surprising that during ischaemia the greatest effects were seen in epicardial circumferential velocities. Further studies, including closed chest or closed pericardium models, are needed to draw firm conclusions on which direction is best for epicardial velocity measurements in the clinical setting. Specificity for the accelerometer in detecting ischaemia It is of major importance to discriminate myocardial ischaemia from other causes of altered myocardial function. A main finding in our study was that although interventions on global myocardial function caused large haemodynamic changes, the effects on circumferential velocities were lesser or different to those caused by ischaemia. This indicates that the accelerometer had the ability to distinguish ischaemia from changes in load and contractility. The high negative predictive values may be of great clinical importance, since ischaemia can be excluded as a reason for clinical alterations. The positive predictive values to detect ischaemia should be interpreted with caution. In this experimental study, a small number of s (n¼12) were performed compared with the non-ischaemic situation (n¼18). Therefore, the few data falsely classified as ischaemia reduced the positive predictive value considerably. In the non-ischaemic region, significant changes in accelerometer systolic velocities were seen in all directions, while myocardial strain did not change in any direction. This denotes that the epicardial velocities were affected by a tethering effect and that accelerometers did not precisely localize the myocardial area affected by ischaemia. This is in accordance with previous observations The number of sensors needed to be used to monitor ischaemia in the presence of three-vessel disease remains unknown; however, a technique that is affected by tethering might be beneficial. By applying a one-axis accelerometer, Theres and colleagues 7 showed that apex motion was affected regardless of which main coronary artery was occluded. Thus, because of the effects of tethering, one accelerometer on the LV apical anterior region may be sufficient for detecting large degrees of ischaemia even in other heart regions. A close correlation between systolic acceleration and LV dp/dt max during the changes in inotropy has been reported in several previous studies. 922 This was the first study to demonstrate that systolic velocity generated from an accelerometer could be a marker of global cardiac function. Strong correlations were demonstrated between accelerometer circumferential systolic velocity and cardiac output and LV dp/dt during interventions affecting global LV function. These results together with the high sensitivity and specificity of the accelerometer in detecting ischaemia indicate that this technique may be utilized both to detect myocardial ischaemia and to assess global LV function. Limitation Our results were obtained in a small number of anaesthetized, mechanically ventilated, and acutely instrumented pigs with an open chest and an open pericardium. For the detection of ischaemia using accelerometry in patients with coronary artery disease, cut-off values, sensitivity, specificity, and predictive values may be different. Studies in patients require the use of a miniaturized accelerometer with dimensions smaller than our prototype. The accelerometer should preferably be incorporated into temporary pacemaker wires which could be attached to the epicardium during surgery and withdrawn after operation, when continued echocardiography monitoring is cumbersome. Miniaturization of the accelerometer is technically feasible and work is in progress. In the clinical setting, automated analysis of the accelerometer velocity curves has to be performed before continuous real-time detection of myocardial ischaemia could 36

9 Detection of myocardial ischaemia by accelerometers be an option. Under these conditions monitoring with accelerometers may be advantageous compared with echocardiography, in particular in the postoperative period. Nevertheless, accelerometers could not replace echocardiography, but may rather be a supplementary screening method for myocardial ischaemia. These issues, together with the unsolved question of how many accelerometers are needed for monitoring all myocardial regions, require further investigation. Conclusion This study showed that an epicardial prototype three-axis accelerometer detected regional myocardial ischaemia, demonstrated by marked changes in accelerometer circumferential velocities. Our results indicate that the accelerometer could distinguish ischaemia from interventions altering global myocardial function. This technique may be used for continuous real-time monitoring of myocardial ischaemia during and after cardiac surgery. However, further studies are needed in patients, during open chest surgery and after the end of surgery. Acknowledgements We thank Halfdan Ihlen for valuable help with design of the study and in preparing the manuscript. Funding Funding for this work was provided by The Regional Health Authorities of Southern Norway, The Research Council of Norway, and The Norwegian Council of Cardiovascular Diseases. References 1 Crescenzi G, Bove T, Pappalardo F, et al. Clinical significance of a new Q wave after cardiac surgery. Eur J Cardiothorac Surg 24; 25: Jain U, Laflamme CJ, Aggarwal A, et al. Electrocardiographic and hemodynamic changes and their association with myocardial infarction during coronary artery bypass surgery. A multicenter study. Multicenter Study of Perioperative Ischemia (McSPI) Research Group. Anesthesiology 1997; 86: Comunale ME, Body SC, Ley C, et al. The concordance of intraoperative left ventricular wall-motion abnormalities and electrocardiographic S-T segment changes: association with outcome after coronary revascularization. Multicenter Study of Perioperative Ischemia (McSPI) Research Group. Anesthesiology 1998; 88: Skarvan K, Filipovic M, Wang J, Brett W, Seeberger M. Use of myocardial tissue Doppler imaging for intraoperative monitoring of left ventricular function. Br J Anaesth 23; 91: Kneeshaw JD. Transoesophageal echocardiography (TOE) in the operating room. Br J Anaesth 26; 97: Elle OJ, Halvorsen S, Gulbrandsen MG, et al. Early recognition of regional cardiac ischemia using a 3-axis accelerometer sensor. Physiol Meas 25; 26: Theres HP, Kaiser DR, Nelson SD, et al. Detection of acute myocardial ischemia during percutaneous transluminal coronary angioplasty by endocardial acceleration. Pacing Clin Electrophysiol 24; 27: Wood JC, Festen MP, Lim MJ, Buda AJ, Barry DT. Regional effects of myocardial ischemia on epicardially recorded canine first heart sounds. J Appl Physiol 1994; 76: Rickards AF, Bombardini T, Corbucci G, Plicchi G. An implantable intracardiac accelerometer for monitoring myocardial contractility. The Multicenter PEA Study Group. Pacing Clin Electrophysiol 1996; 19: Hoff L, Elle OJ, Grimnes MJ, Halvorsen S, Alker HJ, Fosse E. Measurements of heart motion using accelerometers. Conf Proc IEEE Eng Med Biol Soc 24; 3: Edvardsen T, Urheim S, Skulstad H, Steine K, Ihlen H, Smiseth OA. Quantification of left ventricular systolic function by tissue Doppler echocardiography added value of measuring pre- and postejection velocities in ischemic myocardium. Circulation 22; 15: Takayama M, Norris RM, Brown MA, Armiger LC, Rivers JT, White HD. Postsystolic shortening of acutely ischemic canine myocardium predicts early and late recovery of function after coronary artery reperfusion. Circulation 1988; 78: Amundsen BH, Helle-Valle T, Edvardsen T, et al. Noninvasive myocardial strain measurement by speckle tracking echocardiography: validation against sonomicrometry and tagged magnetic resonance imaging. J Am Coll Cardiol 26; 47: Gjesdal O, Hopp E, Vartdal T, et al. Global longitudinal strain measured by two-dimensional speckle tracking echocardiography is closely related to myocardial infarct size in chronic ischaemic heart disease. Clin Sci (Lond) 27; 113: Helle-Valle T, Edvardsen T, Crosby J, et al. New non-invasive method for assessment of LV rotation speckle tracking echocardiography. Circulation 24; 11: Imenes K, Aasmundtveit K, Husa EM, et al. Assembly and packaging of a three-axis micro accelerometer used for detection of heart infarction. Biomed Microdevices 27; 9: Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1: Edvardsen T, Skulstad H, Aakhus S, Urheim S, Ihlen H. Regional myocardial systolic function during acute myocardial ischemia assessed by strain Doppler echocardiography. J Am Coll Cardiol 21; 37: Urheim S, Edvardsen T, Torp H, Angelsen B, Smiseth OA. Myocardial strain by Doppler echocardiography validation of a new method to quantify regional myocardial function. Circulation 2; 12: Gibbons Kroeker CA, Adeeb S, Tyberg JV, Shrive NG. A 2D FE model of the heart demonstrates the role of the pericardium in ventricular deformation. Am J Physiol Heart Circ Physiol 26; 291: H Skulstad H, Andersen K, Edvardsen T, et al. Detection of ischemia and new insight into left ventricular physiology by strain Doppler and tissue velocity imaging: assessment during coronary bypass operation of the beating heart. J Am Soc Echocardiogr 24; 17: Lyseggen E, Rabben SI, Skulstad H, Urheim S, Risoe C, Smiseth OA. Myocardial acceleration during isovolumic contraction: relationship to contractility. Circulation 25; 111: