Cardiac reflections and natural vibrations: Force-frequency relation recording system in the stress echo lab

We enrolled 88 consecutive patients (56 males, 62 ± 13 years) referred for stress echocardiography (49 for exercise, 31 for dipyridamole, 8 for pacing stress). The type of stress was clinically driven for exercise vs. dipyridamole, and by the presence of a permanent pace maker for pacing stress. The characteristics of the study patients are reported in Table 1. The local Ethical Committee approved the study protocol. All the patients gave their written informed consent before entering the study.

All patients met the following inclusion criteria: 1) referred to stress echo for clinically-driven testing; 2) acoustic window of acceptable quality; 3) willingness to enter the study. Exclusion criteria were: 1) unstable angina or recent myocardial infarction; 2) moderate-to-severe aortic stenosis; 3) hemodynamic instability, documentation of life-threatening ventricular arrhythmias (sustained ventricular tachycardia or ventricular fibrillation), atrial fibrillation. From the initially considered population of 92 patients, 4 were excluded for poor acoustic window (n = 2), or refusal to give written informed consent (n = 2).

Graded bicycle semi-supine exercise echo was performed in 49 patients starting at an initial workload of 25 watts lasting for 2 minutes; thereafter the workload was increased stepwise by 25 watts at 2 minutes interval. A 12-lead electrocardiogram and blood pressure determination were performed at baseline and every minute thereafter [16].

Dipyridamole stress echo was performed following the protocol of American Society of Echocardiography [16, 17], using Dipyridamole 0.84 mg/kg in 6' (accelerated protocol) in 31 patients.

Contraindications of using dipyridamole were asthma, hypotension, bradyarrhythmias.

The study population consisted of 8 patients with a permanent pace maker. The pacing protocol was accelerated (with a 10-beat increment every 60 s) until one of the following criteria was reached: 1 – 85% of maximal heart rate (age-corrected: 220 – age for men, 200 – age for women); or 2 – PM maximal programmable heart rate (which varied widely, according to the model of PM, up to 170 bpm during stress). Stimulation was performed, wherever possible, in atrial stimulation mode, or dual-chamber (DDD) pacing to have normal contraction sequence. In the VVI-implanted patients, ventricular stimulation mode was used [10].

The diagnostic end-points for all types of stress were: the development of obvious echocardiography positivity, obvious alterations of ECG (ST segment shift >3 mm). The exam was also stopped in case of limiting subjective side effects or hypertension (systolic pressure >220 mmHg, diastolic pressure >120 mmHg), hypotension (relative or absolute) with decrease of the blood pressure >30 mmHg, supraventricular arrhythmias (supraventricular tachycardia, atrial fibrillation), ventricular arrhythmias (ventricular tachycardia, frequent and polymorphous ventricular beats); limiting dyspnoea, or maximal predicted heart rate in the absence of ischemia [16].

All patients underwent transthoracic echocardiography at baseline and during stress.

Left ventricular end-diastolic and end-systolic volumes were measured from apical four- and two-chamber view, by an experienced observer using the biplane Simpson-method [18]. Only representative cycles with optimal endocardial visualization were measured and the average of three measurements was taken. The endocardial border was traced, excluding the papillary muscles. The frame captured at the R wave of the ECG was considered to be the end-diastolic frame, and the frame with the smallest left ventricular cavity the end systolic frame. Images were acquired at baseline and at each increase in heart rate of 10 beats during stress.

The Wall Motion Score Index (WMSI) was calculated in each patient at baseline and peak stress, according to the recommendations of the American Society of Echocardiography from 1 = normal-hyperkinetic to 4 = dyskinetic in a 17 segment model of the left ventricle [16, 19].

One nurse recorded blood pressures at rest and during each individual study. The blood pressure recording was made using a sphygmomanometer and the diaphragm of a standard stethoscope. Systolic and diastolic blood pressure was obtained in the right arm. During exercise test, blood pressure recording was obtained with patient lying in a left rotated semi supine position and instructed to hand grip to the left support with their left hand. Patients have been told to let their right hand go limp when blood pressure was measured.

In order to build the force-frequency relation, the force was determined at each step as the ratio of the systolic pressure (cuff sphygmomanometer)/end-systolic volume index (biplane Simpson rule/body surface area). At each heart rate step three cardiac cycles were analyzed and the ESV was calculated as the mean value. The FFR was built off line. The slope of the relationship was calculated as the SP/ESV (Systolic Pressure/End-Systolic Volume) index increase (Δ SP/ESV index from baseline to peak stress) (Figure 1 and Figure 5). The FFR was defined up-sloping when peak exercise SP/ESV index was higher than baseline and intermediate stress values; biphasic, with an initial up-sloping followed by a later down-sloping trend, when peak exercise systolic pressure/end-systolic volume index was lower than intermediate stress values [9‐13]; flat or negative, when peak exercise systolic pressure/end-systolic volume index was equal to or lower than baseline stress values.

In all, ventricular arterial coupling was indexed by the ratio of left ventricular systolic elastance index (systolic pressure/end-systolic volume index) to arterial elastance (Ea, ratio of end-systolic pressure by stroke volume). Echocardiography (for ESV and stroke volume) and cuff sphygmomanometer (systolic pressure, multiplied x 0.90 to obtain end-systolic pressure) provided the raw measurements.

Because stroke volume (and input impedance) varies directly with body size, arterial elastance was adjusted for body surface area (EaI) to better reflect differences in arterial properties with age and between the genders adjusted for differences in body size [20, 21]. Of note ventricular-arterial coupling is ventricular elastance/arterial elastance, which can further be described as: end-systolic pressure/end-systolic LV volume divided by end-systolic pressure/stroke volume: the pressure terms in the numerator and the denominator cancel out, and ventricular-arterial coupling equals to stroke volume/end-systolic volume.

SVR were calculated according to the traditional formula:

where 5 is an approximation of the right atrial pressure and MAP is mean arterial pressure.

Systemic arterial compliance (C) was calculated as stroke volume index/systemic arterial pulse pressure; where pulse pressure = systolic blood pressure - diastolic blood pressure [22].

Isovolumic myocardium contractions generate vibrations which have audible components that are responsible for the first heart sound [7, 8]. These vibrations can be measured with modern accelerometer based technology. These devices are small in size and can be easily used to build a cutaneous sensor (Figure 2).

The transcutaneous force sensor is based on a linear accelerometer from STMicroelectronics (LIS3). The device includes in one single package a MEMS sensor that measures a capacitance variation in response to movement or inclination and a factory trimmed interface chip that converts the capacitance variations into analog signal proportional to the motion. The device has a full scale of ±2·g (g = 9.8 m/s²) with a resolution of 0.0005·g. We housed the device in a small case (Figure 3) which was positioned in the mid-sternal precordial region and was fastened by a solid gel ECG electrode. The acceleration signal was converted to digital and recorded by a laptop PC, together with an ECG signal. The system is also provided with a user interface that shows both the acceleration and the ECG signals while the acquisition is in progress. The data were analyzed by using software developed in Matlab (The MathWorks, Inc). A QRS detection algorithm was used to automatically locate the beginning of the isovolumic ventricular contractions [23]. The amplitude of the vibration due to isovolumic myocardium contraction was then obtained to record systolic force for each cardiac beat [24, 25]. While echocardiography uses artificially generated cardiac reflections, the accelerometer simply records naturally generated heart vibrations, which audible components in the isovolumic (preejection) contraction period give rise to the first heart sound. Non myocardial noising vibrations (skeletal muscles, body movements, breathing) were eliminated by frequency filtering (Figure 4, 5).

The curve of force variation as a function of heart rate was finally computed as the increment with respect to the resting force value. All the parameters were acquired as instantaneous values at baseline and during stress; mobile mean was utilized to assess baseline value (1 minute recording), at each incremental stress test, at peak test, and during recovery. Baseline, peak stress, peak-rest difference as absolute value, and delta % rest-peak stress values were computed. The force-frequency relation was defined up-sloping when the peak stress force was higher than baseline and intermediate stress values; biphasic, with an initial up-sloping followed by a later down-sloping trend, when the peak stress force was lower than intermediate stress values; flat or negative, when the peak stress force was equal to or lower than baseline stress values [6, 14]. The critical heart rate (or optimum stimulation frequency) was defined as the heart rate at which the force reaches the maximum value during progressive increase in heart rate; in biphasic pattern, the critical heart rate is the heart rate beyond which the force has declined by 5%; in negative pattern the critical heart rate is the starting heart rate.

SPSS 11 for Windows was utilized for statistical analysis. The statistical analyses included descriptive statistics (frequency and percentage of categorical variables and mean and standard deviation of continuous variables). Pearson chi-square with Fisher's exact test for categorical variables and the Mann-Whitney test for continuous variables for inter-group comparisons were performed to confirm significance (using Monte Carlo method for small sample comparisons).

The one-way ANOVA was used to compare continuous variables between groups; when homogeneity of variance was not present, the Kruskal-Wallis test for nonparametric independent samples was used. Intergroup comparison was performed with Scheffe and Tamhane post hoc tests, respectively.

Relations between variables were assessed using linear regression analysis and Pearson's correlation coefficient.

Analysis of variance with Fisher's post hoc pair wise multiple comparisons was used to assess the significance of intragroup repeated measures. The best predictors of the sensor built force-frequency relation slope and shape were assessed. Receiver operator characteristic (ROC) curve analysis was used to determine potentially useful threshold values of the sensor built FFR as cut-off for normal vs. abnormal contractile reserve (as assessed by the standard echo method: Δ Peak-rest SP/ESV ≥ 4 vs. <4 in exercise, positive vs. negative FFR in Dipyridamole stress). Contingency tables were constructed to assess the sensitivity and the specificity of the new operator independent method to assess the normal threshold for contractile reserve.

Technically adequate images were obtained in all patients at baseline (by selection) and during stress.

A stable, reproducible, and consistent isovolumic systolic force signal was obtained in all patients. This high amplitude vibration is an expression of the tension wave front produced during initial activation of the heart in the isovolumic contraction phase [23]. Baseline force value had an ample range (from 4 to 30 g * 10^-3). Cardiac vibrations propagate as mechanical shear waves, and the intervening viscoelastic thoracic tissue attenuates the higher frequencies and introduces a variable propagation delay [26, 27]. The absolute force value in the single patient can be related to the transthoracic propagation of cardiac vibrations. In fact, when measured epicardially, preejection cardiac vibrations are up to 10 times more powerful than when measured on the chest. In our experimental model, changes of the isovolumic systolic force were linearly related to the echo SP/ESV index changes. Equally, normal vs. abnormal echo built FFR was identified by the sensor built FFR. Attempts have been made by other authors in animals and in humans to assess the usefulness of myocardial preejection vibration amplitude (which audible components give rise to the first heart sound) to quantify left ventricular contractility. Wood et al [28], studied the regional effects of myocardial ischemia on the epicardially recorded first heart sound in dogs, demonstrating that the amplitude of cardiac vibrations in the isovolumic systole are sensitive to global left ventricular contractility and independent of the site of application of the sensor (normal contracting or ischemic non-contracting area). Sakamoto et al [8] demonstrated that variations of the first heart sound amplitude are correlated to variations of the isovolumic maximum rise of LV pressure development (max. LV dP/dt).

However, to our knowledge, this is the first work in which the isovolumic systolic myocardial force vibration was utilized to build the force-frequency relation and in which, for the first time, a comparison with the echo built FFR has been made [7]. Since FFR slope and shape are key determinants of clinical outcome in diseased hearts, [11‐13] FFR should be extensively monitored with this novel noninvasive operator-independent method. For sake of clarity, the conceptual and methodological differences between the FFR and the sound sensor system are contrasted in Table 4.

Exercise evaluates the effect of heart rate increase and inotropic reserve due to adrenergic stimulation during exercise, rather than pure heart rate effect [1, 29, 30].

The basic property of the force-frequency effect to progressively enhance myocardial contractility as heart rate increases is augmented at each level of increasing adrenergic stimulation as occurs in exercise by catecholamine spillover. The sensor built FFR fulfilled recording of this typical pattern, both as absolute values, and for the capability to distinguish normal vs. abnormal contractile reserve, as confirmed by the standard echo built FFR cut-off (Figure 6).

Dipyridamole has the well known coronary vasodilator effects mediated by the inhibition of adenosine cellular transport, eventually leading to extra cellular adenosine accumulation and steal phenomena [17]. There is a mild catecholamine release that is responsible for the mild inotropic effect of the drug (Figure 8). Simultaneously, the "pure" heart rate dependent increase in contractility is blunted due to the low heart rates increase during dipyridamole stress. These data are consistent with our findings; where there was a more pronounced and prolonged increase in the force signal in the exercise group vs. Dip group (Table 2). Despite the low number of patients, the pacing group showed intermediate FFR slope between exercise and dipyridamole: catecholamine levels are generally unchanged during the increase in heart rate produced by pacing [10]. Therefore, at least theoretically, pacemaker stress echo is more suitable to assess FFR rather than inotropic reserve.

The assessment of force-frequency relationship is a theoretically robust approach for evaluating left ventricular contractility, which has been deployed clinically using invasive, complex and technically demanding methods [3‐6]. Previously, we proposed a non-invasive echocardiographic method [9‐13] to assess the changes in contractility during exercise echo. FFR recording is feasible with a single precordial sensor fastened by a standard solid gel ECG electrode, in a totally automatic, operator independent fashion. FFR recording with a non-invasive sensor offers a new chance to monitor indexes of LV systolic function, especially in failing hearts and/or exposed to drugs [31].

In diseased hearts the force-frequency relation is altered. As the heart fails, there is a change in the gene expression from the normal adult pattern to that of fetal life with an inversion of the normal positive slope of the force-frequency relation: systolic calcium release and diastolic calcium reuptake process is lowered at the basal state and, instead of accelerating for increasing heart rates, slows down [12]. The force-frequency relation of these diseased hearts exhibits a flat or negative slope at contraction frequencies above about 100 bpm [14, 15] (Figure 10). The heart rate at which the contractile force begins to decline diminishes progressively for known pathologies such as ischemic cardiomyopathy, diabetic cardiomyopathy, mitral regurgitation and dilated cardiomyopathy. The optimum contraction rate, i.e. the rate corresponding to the strongest contractile force, varies for each pathology, for each different stage of the illness, and for each patient [23]. Different therapies are known to treat chronic heart failure, however the response must be tailored to each patient. For example, therapy with beta blockers has proved effective in cardiomegaly regression and in improving the myocardial function in patients suffering from dilated cardiomyopathy, however patients do not respond uniformly to this therapy. The bradycardic action of beta blockers reduces the number of daily working phases of the heart in the negative part of the force-frequency relationship. As the flattening and the descending limb of the force-frequency curve appears at different frequencies in the different types of cardiac decompensation and in different patients, the effectiveness of the action of beta blockers is variable. To optimize chronic heart failure therapy, it therefore appears essential to identify in each patient the ascending part of the force-frequency curve and the specific rate which, when exceeded, initiates the flattening and the descending part, in order to optimize individual therapy for chronic heart failure [6]. Monitoring the patient's conditions with this new sensor built FFR not only enables his condition to be verified in terms of identifying an abnormal situation not in itself critical but predictive of a future worsening of chronic heart failure, but also it enables the effectiveness of therapies and their influence on the patient's condition under conditions of normality to be verified [31] (Figure 11). As distribution of the daily activities of the patient is variable, during the day the patient experiences, depending on the activities normally performed, periods in which his force-frequency relationship is of positive slope and periods in which his force-frequency relationship is of negative slope or flat. Known check up systems do not enable therapy effectiveness to be verified other than by occasional checks on the patient under rest conditions. By monitoring in accordance with this new sensor, not only can positive or negative decompensation changes as a result of certain events be determined, but also the variation in the force-frequency curve over twenty-four hours, days or weeks can be evaluated. A three-dimensional diagram could be determined which for each heart rate not only indicates the instantaneous force value but also enables the variation of said value with time to be monitored. (Figure 12). If the curve of force variation at a determined heart rate during the course of the day is reproduced as a two-dimensional diagram, it gives a continuous indication of the progress of decompensation situations in response to the therapies adopted.

The patient's number and the incoming disease were different in the exercise, dipyridamole and pacing stress groups. Controls were present only in the exercise group: it isn't possible to have physiological and hemodynamic comparisons between stress groups. However, aim of this study, was not to assess physiological patterns and differences between type of stress, that are well known, but simply assess the feasibility of sensor isovolumic contraction force recording during stress echocardiography, and to compare the standard operator-dependent echo vs. this sensor operator-independent built force-frequency relation during stress.

Because only non-invasive measurements were available, systolic cuff pressure was used as a surrogate for end-systolic pressure. The non-invasive results tend to overestimate the actual ratio, in large part because the brachial arterial pressure reflects an amplified LV pressure wave secondary to the rheological properties of the arterial tree, and this difference may lead to overestimation of left ventricular pressure, specially if younger subjects are studied, in which blood pressure tends to be overestimated [32]. However, the peak systolic pressure/end systolic volume ratio is a reproducible method of assessing left ventricular performance during exercise [33]. Several experimental studies in animals [34] and in human beings [35] have shown that use of peak systolic pressure yielded nearly identical pressure-volume relations.

The estimation of the LV systolic elastance index by SP/ESV index assumes V₀ = 0. The volume axis intercept of the end-systolic pressure/volume relationship V₀ is generally lower in normal subjects than in CHF patients, and may also differ among DCM patients. Nevertheless, SP/ESV index is generally considered as a valid approximation of LV end systolic elastance [36]. This may not hold true during exercise if V₀ changes. It has been reported that V₀ is not appreciably altered by inotropic stimulation [37] or changes in loading conditions: Maughan et al [38] showed that V₀ changed by only 5 mL on average in response to changes in peripheral vascular resistances of 50 to 200% of normal; Little and Cheng [39] observed no significant change in V₀ during treadmill exercise in dogs. It is likely, however, that V₀ changes during exercise but this is impossible to assess at such levels of exercise in man.

Since end-systolic elastance (Ees), expressing the slope of the in-vivo, end-systolic ventricular pressure vs chamber volume relation, is the most "foolproof" window into in vivo myocardial contractility, Ees should be measured at each heart rate step increase, as made by Liu and coworkers [3]. But measuring Ees for increasing heart rates is impractical: increasing heart rates obtained with temporarily pacing has to be adjunct to the LV conductance catheter, the LV pressure catheter, the vena cava balloon, and to afterload changes. However, at least cath lab animal studies should be done to validate the noninvasive, operator independent, sensor builded FFR.

Medication differences may have had an impact, blunting the force-frequency relation and the diastolic blood pressure. However, this variable could not be controlled for ethical and practical reasons. In addition, the index is relatively independent of afterload changes, since for a given level of contractility – blood pressure reduction should be accompanied by smaller end-systolic volume values.