Recovery of Systolic and Diastolic Left Ventricular Function Early after Cardiopulmonary Bypass

This study was designed to document recovery of LV function early after separation from CPB and to establish possible physiologic mechanisms involved in regulating LV function during this period.

The study consisted of two protocols. In a first protocol (protocol 1), time course of recovery of LV function early after separation from CPB was documented in patients undergoing coronary artery surgery using two different surgical and myocardial protective techniques. One subset of patients was operated on using intermittent aortic cross-clamping under cardioprotection with the nucleoside transport inhibitor lidoflazine (Group A) and the other subset of patients was operated on using continuous aortic cross-clamping under cardioprotection with Bretschneider cardioplegic solution (Group B). In a second protocol (protocol 2), the time course of recovery of LV function early after separation from CPB was documented in patients with a different regimen of preload management before separation from CPB (Group C).

The study was attempted in 44 patients scheduled for elective coronary bypass surgery. The study was approved by the Institutional Ethical Committee and informed consent was obtained in all patients. Patients with an ejection fraction of more than 40% or with an LV end-diastolic pressure of less than 15 mmHg on preoperative hemodynamic evaluation were considered. Patients with unstable angina or undergoing repeated coronary surgery, concurrent valve repair, or aneurysm resection were excluded. At the time of the study patients were hemodynamically stable.

Premedication consisted of 2 micro gram/kg intramuscular glycopyrolate (Robinul), 30 micro gram/kg droperidol, and 1 micro gram/kg fentanyl given 90 min before surgery. In the operating room, patients received routine monitoring, including five-lead electrocardiography, radial and pulmonary artery pressure, pulse oxymetry, capnography, and blood and urine bladder temperature. After preoxygenation, anesthesia was induced with 20 micro gram/kg fentanyl, 0.1 mg/kg diazepam, and 0.1 mg/kg pancuronium bromide. An additional dose of 30 micro gram/kg fentanyl was administered before sternotomy. Patients' lungs were ventilated with FiO²was 0.5; when necessary, 0.2% to 0.4% isoflurane was added to the air-oxygen mixture. All patients routinely received 2 g methylprednisolone after induction of anesthesia and 2.10⁶KIU aprotinin in the priming fluid of the extracorporeal circuit.

Echocardiographic data were acquired using a biplane 5-MHz esophageal ultrasound probe (Aloka UST-5233-S; Tokyo, Japan) connected to a SSD-830 Aloka echocardiographic unit. Short-axis transgastric incidences were selected for analysis. The midpapillary muscle level was taken as an anatomic landmark and the probe was positioned to obtain the image with the most circular overall geometry with uniform wall thickness. This plane was selected because this procedure was acceptable during surgical procedures and because earlier studies have shown that this cross-sectional area allowed fair estimation of LV volume. [12-15] 

Standard median sternotomy and pericardiotomy were performed and the heart was prepared for coronary artery bypass grafting. The aorta was cannulated and epicardial pacemaker wires were attached to the right atrium and right ventricle. Then a fluid-filled catheter with a broad internal lumen and a rigid wall (Cavafix; 18 G, 45 cm long, Braun Melsungen AG, Melsungen, Germany) was positioned in the LV cavity through the apical dimple to record LV pressure. The catheter was connected through 120-cm high-pressure tubing (Viggo-Spectramed, Swindon, UK) to a pressure transducer (M1006A; Hewlett Packard, Danmark). All transducers were calibrated to zero to the air at the beginning of the protocol. Zeroes were checked before each set of measurements. Dynamic response (natural frequency and damping coefficient) of the catheter-transducer system was evaluated using the flush method described by Gardner [16]before and after each experiment. For the LV pressure measurement system, mean frequency response was 32 Hz, with a mean damping coefficient of 0.41. All individual measurements were within the adequate range of dynamic response requirements for direct pressure measurements. The output signal of the pressure transducer system was digitally recorded together with the electrocardiographic signals (Codas; DataQ, Akron, OH).

Depending on the surgeon's preference, two types of surgical technique and cardiac protection were used during coronary artery bypass grafting. One surgeon (I.E.R.) performed coronary bypass surgery using intermittent cross-clamping under cardioprotection with the nucleoside transport inhibitor lidoflazine (0.5 mg/ml concentration; Janssen Pharmaceuticals, Beerse, Belgium). This technique involved intravenous administration of 1 mg/kg lidoflazine before the start of CPB. Distal anastomoses were performed under aortic cross-clamping. After each distal anastomosis, the aortic cross-clamp was released and a reperfusion period of one half the period of ischemia was allowed before the next distal anastomosis was started. The other surgeon (L.R.H.) performed coronary artery bypass surgery after coronary perfusion with cold (4 degrees Celsius) Bretschneider cardioplegic solution without blood (composition per 1,000 ml: 15 mM NaCl; 9 mM KCl; 1 mM K hydrogen-2-oxoglutarate; 4 mM MgCl-6 H²O; 0.015 mM CaCl-2 H²O; 18 mM histidine HCl-H²O; 180 mM histidine; 2 mM tryptophane; and 30 mM mannitol). Patients were randomly assigned to one of the two collaborating surgeons into two equal groups.

Venous drainage during CPB was obtained with a two-stage venous cannula inserted into the right atrium. A ventricular sump was inserted into the left ventricle through the right superior pulmonary vein. Perfusion flow on CPB was 2.4 l [centered dot] m sup -2 [centered dot] min sup -1 in nonpulsatile mode. Patient body temperatures were cooled to a bladder temperature of 28 degrees Celsius. In all patients the left internal thoracic artery was used in addition to one or more saphenous vein grafts.

The heart was paced for the duration of the protocol at a fixed rate of 90 bpm in atrioventricular sequential mode, with an atrioventricular interval of 150 ms. No vasoactive or inotropic drugs were allowed during the study.

Measurements were obtained with the ventilation suspended at end-expiration. Measurements consisted of high-speed recordings of digitized electrocardiographic and LV pressure tracings and echocardiographic images on videotape. These data were recorded during a progressive increase of systolic and diastolic LV pressures obtained by tilting the caudal part of the surgical table by 45 degrees, thus elevating the patients' legs. Measurements of pressure and dimension data were synchronized by sending a synchronized electronic signal at the beginning and at the end of the recording. Care was taken to have at least 15 consecutive beats for analysis. After recording, the surgical table was returned to the horizontal position.

Baseline measurements (before CPB) were obtained when the surgical preparation was completed. After this measurement, the ventricular catheter was removed, the venous cannula inserted, and CPB initiated. After the surgical procedure, reperfusion of the heart (reperfusion time was set at 50% of the aortic cross-clamp time in all patients), and rewarming to a bladder temperature of 35 degrees Celsius, the ventricular pressure catheter was repositioned in the LV cavity and patients were separated from CPB. The heart was paced again and filled until a pulmonary capillary wedge pressure of 13 to 15 mmHg or a central venous pressure of 8 to 10 mmHg was achieved. When, after termination of CPB, cardiac index exceeded 2 l [centered dot] m sup -2 [centered dot] min sup -1 with a systolic systemic arterial pressure greater than 75 mmHg or a mean arterial pressure greater than 50 mmHg, the venous cannula was withdrawn and the experimental protocol was resumed. Measurements were obtained immediately after stabilization following venous decanulation (time 0) and after 5 (time 5), 10 (time 10), and 15 (time 15) min.

Adequate data for analysis were obtained in 24 of the 30 patients. In one patient in Group A, atrioventricular sequential pacing was not possible in the pre-CPB period because of frequent supraventricular extrasystoles. In one patient in Group B, no high-quality transgastric echocardiographic images suitable for analysis were obtained. Two patients were excluded because they required nitroglycerin before and immediately after CPB to control their hemodynamic parameters (one patient in each group). Two other patients were excluded because they needed inotropic support with a continuous infusion of 5 micro gram [centered dot] kg sup -1 [centered dot] min sup -1 dobutamine (one in each group).

To evaluate possible underlying physiologic mechanisms that regulate LV function early after separation from CPB, a second protocol was started in which a different preload management was followed. These patients were also randomly assigned in two equal groups to the two collaborating surgeons. Adequate data for analysis were obtained in 12 of 14 patients (Group C). Two patients were excluded because no high-quality transgastric echocardiographic images suitable for analysis were obtained. In the patients enrolled in protocol 2, the separation procedure was adapted by using a different preload management at the end of CPB. The heart was filled until a pulmonary capillary wedge pressure of 13 to 15 mmHg or a central venous pressure of 8 to 10 mmHg was achieved 10 min before anticipated separation from CPB. Total reperfusion time in these patients was also 50% of the aortic cross-clamp time, which means that reperfusion time was identical in both groups. The heart was paced and allowed to eject while flow on CPB was slightly reduced (maximum, 15%) to prevent emptying of the venous reservoir of the extracorporeal circuit. This situation was kept stable for 10 min and then separation from CPB was initiated. For both protocols, separation procedures were standardized to take 3 min in all patients.

Surgery was uncomplicated in all patients studied and no patients developed a perioperative infarction based on the appearance of elevated creatine kinase-MB > 20 or new Q waves on an electrocardiogram. Table 1summarizes the clinical demographic and intraoperative data of the patients included in the study.

End-diastolic pressure was timed at the peak of the R wave on the electrocardiogram. The first derivative of LV pressure (dP/dt) was calculated using digitized data with a seven-point smoothing algorithm.

Echocardiographic data were recorded on VHS videotape (Panasonic, AG-7330, Matsashuti, Japan) and analyzed off-line blinded from the LV pressure data. End-diastole was measured at the point of maximal LV cavity area, whereas end-systole was measured at the point of minimal LV cavity area. If this was not readily available, end-diastole and end-systole were measured when the electrocardiograph-gated freeze-frame analysis of echocardiographic images corresponded to the peak of the R wave and to the end of the T wave, respectively. Care was taken that end-diastole and end-systole were timed in the same way in the consecutive measurements of each patient. Endocardial borders were manually outlined from the video screen with a trackball. Images were evaluated according to the position and quality scores proposed by London and associates. [17]Only patients with a position score of 1 (optimal short-axis orientation) and a quality score of 3 (good endocardial and epicardial resolution) were included for echocardiographic analysis. The papillary muscles were excluded when the endocardial border was traced.

Left ventricular dimensions were measured and calculated using a computer program (Echo-com; PPG Hellige GmbH, Freiburg, Germany). The measured cross-sectional area corresponded to pi [centered dot] r². Assuming a spherical model, ventricular cavity volume (V) was calculated with the formula: V = (4 pi/3) [centered dot] r³. The measured cross-sectional cavity areas were reported in the tables and derived cavity volumes were used to compute ventricular systolic elastance and diastolic stiffness. Regional wall motion abnormalities may influence the accuracy of calculating ventricular volumes based on a spherical model using a radius calculated from cross-sectional areas measured by echocardiography. In our study, none of the patients exhibited severe wall motion abnormalities such as dyskinetic or akinetic segments. All patients exhibiting akinetic or dyskinetic regional segment wall behavior also had ejection fractions less than 35% and were already excluded from the protocol for this reason.

Ten to 15 consecutive beats were taken for LV pressure and dimension analysis. End-diastolic pressure-dimension relations were constructed for each set of measurements. Passive properties of the ventricle were described by fitting an exponential equation through these points using the three-constant equation that allows LV pressure to decay to a natural asymptote [18]: Equation 1where P and V represent the corresponding end-diastolic pressure and volume, Kc corresponds to the stiffness constant, and A and B are empirical constants.

Systolic performance was assessed by evaluating the end-systolic pressure/volume relation. End-systolic volume was taken as minimal systolic area and end-systolic pressure was defined as pressure at dP/dt^min. The corresponding systolic pressure and volume data were fit by linear least-squares analysis to the following equation: Equation 2where P = LV pressure, Ees = the slope of the systolic pressure - volume relation, V = LV systolic volume, and VO = the volume intercept of the systolic pressure - volume relation. Sample correlation coefficients of the end-diastolic and the end-systolic pressure-volume relationship yielded values of r > 0.92 in all patients.

In addition, mean arterial pressure, mean pulmonary arterial pressure, right atrial pressure, and cardiac index were measured using the thermodilution technique. These variables were used to calculate systemic vascular resistance with Equation 3:

Left ventricular stroke work index were calculated using Equation 4: where LVEDP is left ventricular end-diastolic pressure and SI represents the stroke index.

Data for protocol 1 were analyzed using a two-factor analysis of variance for repeated measurements to assess the effects of time in both types of surgical procedure. Interaction analysis showed whether the type of surgical procedure had a significant effect on the repeated measurements. Post-test hypothesis used the Scheffe F-test when appropriate. Measurements in both surgical techniques appeared to be identical for all parameters measured. Data from protocol 1 thus were pooled and compared with data from protocol 2 using a two-factor analysis of variance for repeated measurements to assess the effects of time in both types of preload management before separation from CPB. Interaction analysis showed whether the preload management regimen had a significant effect on the repeated measurements. Post-test hypothesis used the Scheffe F-test when appropriate. Pre- and intraoperative data between both groups were compared using Fischer's exact test. Data were reported as means +/- 1 SEM. Statistical significance was accepted at P < 0.01.

Data on LV function in groups A and B, at baseline, time 0, time 5, time 10, and time 15 are summarized in Table 2. Compared with baseline data, LV data at times 0 and 5 showed higher end-diastolic pressure, lower peak LV pressure, and decreased systolic function (dP/dt sub max, area ejection fraction). From time 0 to time 10, a rapid recovery of LV function was observed: end-diastolic pressure decreased, peak LV pressure recovered, and systolic function improved to baseline levels. dP/dt^maxremained lower than at baseline throughout the entire observation period. Left ventricular data at the different points of measurement were similar in both groups. Interaction analysis showed no difference in the effect of time on recovery of LV function for both groups.

Hemodynamic data in groups A and B at baseline and at times 0, 5, 10, and 15 are summarized in Table 3. Compared with baseline, mean arterial pressure was lower after separation from CPB, whereas mean pulmonary arterial pressure and right atrial pressure were increased. These data returned to baseline within 10 min after separation from CPB. Cardiac Index (CI) was also depressed at time 0 and returned progressively toward baseline values within 10 min. Systemic vascular resistance was decreased after CPB and remained lower throughout the observation period. Left ventricular stroke work index was decreased after CPB but also returned to baseline values within 10 min. All hemodynamic data at the different points of measurement were similar in both groups and there was no different effect of time on recovery of hemodynamics for both groups.

Load-independent evaluation of LV function was obtained by analyzing end-systolic and end-diastolic pressure-volume data. A representative set of systolic and diastolic pressure-volume data from patient 13 are shown in Figure 1(left). From time 0 (open squares) to time 10 (open circles), the systolic pressure-volume relation became steeper, whereas the diastolic pressure-volume relation shifted downward, confirming a rapid improvement of systolic and diastolic LV function. These two aspects of cardiac function were quantitatively assessed by the slope Ees of the systolic pressure-volume relation and by the ventricular stiffness constant Kc of the diastolic pressure-volume relation for all observations in both groups (Figure 1, right panel). Ees decreased significantly after CPB but returned to baseline values within 10 min and remained constant thereafter in both groups. Evolution of Ees was similar in both groups. Kc increased after CPB, returned to baseline values within 10 min, and remained also constant thereafter. Evolution of Kc was also similar in both groups. The volume-axis intercept (VOes) at time 0 was decreased compared with baseline and progressively returned to baseline values within 15 min. This change was comparable in both groups (group A: baseline, 15 +/- 11 ml; time 0, -37 +/- 14 ml; time 5, -18 +/- 5 ml; time 10, -7 +/- 6 ml; time 15, 5 +/- 7 ml [significance of differences between repeated measurements: P = 0.008]; group B: baseline, 12 +/- 13 ml; time 0, -41 +/- 16 ml; time 5, -13 +/- 6 ml; time 10, -11 +/- 5 ml; and time 15, 7 +/- 4 ml [P = 0.004]). The pressure-axis intercept (B) was 7 +/- 3 mmHg for group A and 8 +/- 2 mmHg for group B and remained constant at all times of measurement in both groups. Given the limited range of obtained pressure and dimension measurements, no further consideration was given to these intercepts.

Recovery of systolic and diastolic LV function early after separation from CPB was confirmed by evaluating other indices of myocardial function. Figure 2(upper) displays the effects of increasing end-diastolic volume (by elevating patients' legs) on stroke volume for each patient in groups A and B for the different times of measurement. For a comparable increase in end-diastolic volume, increases in stroke volume were significantly higher at time 10 than at time 0 (P = 0.005). A linear relation was found between the increase in end-diastolic volume and the increase in stroke volume for time 0 (r = 0.89; P = 0.004) and for time 10 (r = 0.91; P = 0.007). This was also apparent when effects of increasing end-diastolic volumes on changes in dP/dt^maxwere compared at the different times of measurement. In group A at time 0, leg raising increased end-diastolic volume by 11 +/- 3 ml, with a corresponding change in dP/dt^maxof 20 +/- 7 mmHg/s. At time 10, leg raising increased end-diastolic volume by 12 +/- 6 ml, with a corresponding increase in dP/dt^maxof 110 +/- 12 mmHg/s. This increase was significantly different (P = 0.0048). The same phenomenon was noticed in patients in group B: At time 0, leg raising increased end-diastolic volume by 14 +/- 5 ml, with a corresponding increase in dP/dt^maxof 25 +/- 8 mmHg/s, whereas at time 10, an increase in end-diastolic volume of 13 +/- 6 ml increased dP/dt^maxwith 120 +/- 14 mmHg/s (P = 0.0062).

The recovery of diastolic function was confirmed by evaluating the effects of leg raising on end-diastolic volume and end-diastolic pressure for each patient in groups A and B for the different times of measurement (Figure 2, lower). At time 10, corresponding increases in end-diastolic volume induced lower increases in end-diastolic pressures, confirming recovery of ventricular compliance and diastolic function.

Recovery of ventricular function after separation from CPB was evaluated in 12 patients after a different regimen of preload management. In these patients, the hearts were optimally filled 10 min before anticipated separation time and the heart was allowed to eject. Data obtained in this group of patients (Group C) were compared with the pooled data of groups A and B. Aortic cross-clamp time, CPB time, and total reperfusion time of patients in group C were similar to those in patients in groups A and B.

(Table 4) summarizes left ventricular and hemodynamic data of patients in group C. Baseline data were similar to data for patients in groups A and B. Left ventricular and hemodynamic parameters of patients in Group C remained unchanged at the different times of measurement and were comparable to values obtained at baseline. Only dP/dt^maxand SVR were decreased in a similar way as in groups A and B.

Data on Ees and Kc are displayed in Figure 3(filled squares) and compared with pooled data on Ees and Kc of protocol 1 (filled circles). Ees and Kc remained unchanged at the different times of measurement (P for repeated measurements = 0.491 and 0.631, respectively). Accordingly there was a significant difference between the data obtained in both protocols (Ees: P = 0.002; Kc: P = 0.003). Compared with baseline, systolic and diastolic LV functions in protocol 1 (normal separation procedure) were impaired after separation from CPB but recovered within 10 min. In protocol 2 (modified separation procedure), systolic and diastolic LV function were not impaired after separation from CPB and remained stable throughout the observation period.

Maintenance or improvement of cardiac function is an important goal in the postoperative phase after open heart surgery, because adequate cardiac function ultimately determines patient survival. In patients undergoing coronary bypass grafting, various results on recovery of postoperative ventricular function have been reported. Most studies mention a decrease in ventricular function between 2 and 6 h after operation, with a return to normal within 24 h to 7 days. [1-5]However, studies on recovery of ventricular function after coronary artery surgery do not always mention impaired function after surgery. Reduto and colleagues [19]reported variable effects on LV ejection fraction after surgery. Of the 57 patients studied, 26 had a decrease in LV ejection fraction at the start of the measurements (110 min after the end of surgery), whereas 25 patients exhibited an increase in LV ejection fraction. By the time of the discharge study (7 days later), LV ejection fraction had returned to preoperative values in 24 of 26 and in 23 of 25 patients, respectively. Six of the 57 patients had no change in ejection fraction after operation. Mangano [20]described two distinct patient populations with regard to the ventricular dysfunction-recovery sequence: (1) a moderate biventricular dysfunction immediately after CPB (first measurement made 15 min after the end of CPB) with almost complete recovery within 4 h; and (2) more severe dysfunction with no recovery even after 24 h.

Thus, although various dysfunction-recovery sequence patterns were described, ventricular dysfunction is a common phenomenon after coronary bypass surgery. All of the previously mentioned studies focused on recovery patterns during the first hours and days after surgery, with the earliest reported evaluation of ventricular function made 15 min after the end of CPB. Little is known about the short-term effects of coronary artery surgery on myocardial function. Recently Gorcsan and associates [6]found pressure-area estimates of end-systolic elastance, maximal elastance, and preload-recruitable stroke force to be significantly decreased immediately after the end of CPB. Load-dependent measures of LV function, such as stroke volume, cardiac output, and fractional area change, were unchanged after surgery in these patients. However, that study did not describe recovery patterns during the first minutes after CPB. We evaluated that in the absence of confounding interventions such as inotropic or vasodilating support. A progressive improvement of systolic and diastolic ventricular function was observed within the first 10 min after separation from CPB. For technical reasons (further course of the operation), ventricular pressure data could not be obtained beyond 15 min and therefore our observations were limited to the first 15 min after CPB. Ventricular function at time 15 was not entirely similar to baseline because dP/dt^maxwas still lower at time 10. Based on these observations and the results reported in the literature, it seems that recovery of ventricular function after cardiac surgery is rather complex and may occur in different phases, with one period of early progressive recovery of function within 10 to 15 minutes after separation from CPB and a second period of recovery with return of parameters of ventricular function toward preoperative values only after several hours or even days.

Time course of recovery suggested a mechanism that might be related to restoration of preload. In isolated cardiac muscle, an abrupt increase in diastolic length resulted in an immediate increase in systolic force, which was followed by a secondary slow increase in force in about 10 min. [21-23]These changes in systolic function were also observed in the intact canine heart. [24,25]Based on the similarity of the time dependency of preload-dependent changes in cardiac function and data of the present study, we hypothesized that changes in systolic and diastolic function might be related to time-dependent increase in LV function after acute volume loading immediately before separation from CPB.

We tested this hypothesis by altering the regimen of preload management before the separation procedure. When preload was optimized 10 min before anticipated separation time and the heart was allowed to eject, it appeared that diastolic and systolic function immediately after termination of CPB reached values that were otherwise obtained only after 10 min and that were similar to baseline values. These data therefore suggested that by optimizing preload conditions, 10 min before anticipated separation time, the characteristic time-dependent improvement of systolic and diastolic function after CPB was no longer observed. Instead ventricular function was optimal from time 0 and remained constant during the observation period that followed.

A possible explanation for this phenomenon thus might be the time-dependency of autoregulation of ventricular function by preload. This was characterized by an immediate increase in performance with administration of preload before separation from CPB, followed by a more gradual improvement in ventricular function within the next 10 min. Time-dependent changes in contractility related to preload were described extensively in vitro. [21-23]The acute increase in contractility was attributed to a length-dependent interaction between the contractile proteins and calcium, whereas the secondary, more gradual increase in myocardial function appeared to be related to intracellular calcium fluxes, as shown with aequorin signals. [23] 

More recently, similar findings were reported in the intact canine heart. [24,25]These changes were not due to neural or humoral factors [24]but represented length-dependent alterations in excitation-contraction coupling, which were reviewed earlier. [21,24,25]Our study suggested that this particular inotropic mechanism might be clinically relevant and present with a similar time course in the cardiac patient. As far as we know, changes in diastolic function related to preload were not previously described in vivo, either experimentally or clinically. In isolated cardiac muscle, however, this phenomenon has been described: Cardiac muscle becomes stiffer at lower preload and more compliant at optimal preload each time within 10 min after the acute change in preload was induced. [26] 

The observed differences between both regimens of preload management could not be attributed to differences in study population or pre- and perioperative conditions because there were no differences in demographic and intraoperative parameters. Number of grafts, aortic cross-clamp time, reperfusion time, and total CPB time were similar in all groups. However, possible other mechanisms must be considered in our observations. Pulsatile perfusion to the heart and the systemic vasculature developed 10 min earlier in patients in group C than in patients in groups A and B. Earlier start of pulse pressure may enhance perfusion of peripheral organs and the heart itself, and this could contribute to the functional differences.

Several investigators have studied the effects of pulsatile flow on CPB. Pulsatile perfusion of the nonbeating heart was shown to increase subendocardial perfusion and to maintain a better endocardial-to-epicardial flow ratio. [27,28]Others, however, failed to confirm these observations either in the beating or the fibrillating heart. [29-32]In the last few years, increasing evidence was gathered on the impact of the inflammatory response to CPB on postoperative myocardial function. It was recently suggested that inflammatory mediators released during CPB may adversely affect postoperative ventricular function. [33]Although in our study reperfusion time was similar in both protocols, we cannot preclude the fact that the start of pulsatile flow throughout myocardial vasculature 10 min before anticipated time of separation of CPB might have induced a better washout of such inflammatory mediators. This might result in earlier better ventricular performance. Several drugs have been reported to blunt effects of CPB and reperfusion-induced inflammatory mediators, among these glucocorticoids [34]and aprotinin. [35]All patients in both protocols of the present study routinely received 2 g methylprednisolone after induction of anesthesia and 2.10⁶KIU aprotinin in the priming fluid of the CPB circuit. Thus it is less likely that the effects of the preload management regimen depended on mechanisms that might be influenced by one of these drugs.

We carefully omitted inotropic or vasoactive medication in this study to evaluate recovery of ventricular function without these confounding elements. This means that patients necessitating inotropic or vasoactive support were not included in the present data analysis. This might imply that a selection has occurred of patients with good ventricular and hemodynamic function, including mainly patients in whom rapid recovery of ventricular function might be expected. In this selection of patients it seems reasonable to observe the recovery of cardiac function for 10 min, rather than rushing pharmacologic interventions. This conclusion, however, might not apply for patients whose hearts have poor ventricular function.

A first limitation of our study is the use of a single two-dimensional plane to describe changes in overall LV volume. The accuracy of calculating ventricular volumes based on a spherical model using a radius calculated from cross-sectional areas measured by echocardiography may be impaired in the presence of regional wall motion abnormalities. However, none of the patients we evaluated had dyskinetic or akinetic segments on echocardiographic examination, so this phenomenon did not influence the data.

We used fluid-filled catheters to measure LV pressures. All individual measurements conformed to the dynamic responses required for fluid-filled pressure transducer systems, which indicates that pressure and dP/dt measurements were reliable. The only drawback is that measurement of time intervals might be delayed with respect to measurements by micromanometers. Thus we gave no further consideration to measuring time intervals.

Loading conditions were altered with leg elevation to construct pressure-dimension relations. This implies that the present data cannot simply be compared quantitatively with data obtained by other means of altering loading conditions. In addition, the simple transformation of area data to volume data will provide the smallest possible ventricle (sphere based on short-axis area) and therefore elevate the slope of the end-systolic and end-diastolic pressure-volume relations. Determining end-systolic and end-diastolic pressure-volume relations was based on pressure and dimension data that were obtained for a relatively limited range. This was due to the specificity of the clinical situation, which did not allow us to vary ventricular loading conditions over a wide range of pressures and dimensions. This implies that quantitative determination of axis intercepts may be hazardous. The repetitive observations in the same patient, however, were within the same range of pressure and dimension data, and thus we could compare repetitive measurements for one patient. In addition, recovery of systolic and diastolic ventricular function early after CPB was confirmed by other determinants of ventricular function.

The present data obtained in patients with good LV function who had coronary artery surgery showed impairment of both systolic and diastolic ventricular function immediately after separation from CPB, which recovered progressively within the first 10 min after CPB. Restoring preload to normal 10 min before anticipated separation from CPB and allowing the heart to eject during this period prevented this transient impairment of ventricular function after CPB. The underlying mechanisms of this phenomenon remain to be established definitively. A possible explanation is that this recovery reflected time dependency of regulation of ventricular function by preload.