INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

DIAGNOSIS OF INADEQUATE CARDIAC PRELOAD, a frequent cause of cardiovascular instability in critically ill patients, is difficult to establish at the bedside. Echocardiography permits measurement of right ventricular (RV) and left ventricular (LV) end-diastolic dimensions and is assumed to reflect RV and LV preload. LV end-diastolic area has been shown to be a sensitive index of volume status (4, 5, 8, 14, 16, 18, 19). Moreover, in patients receiving mechanical ventilation, the magnitude of stroke volume (SV) variation over a respiratory cycle has been suggested to detect preload responsiveness (12). As surrogates of SV variation, respiratory variations of systemic arterial pressure (16, 11) or of aortic Doppler velocity (using transesophageal echocardiography) (8) have been reported to predict fluid responsiveness.

The aim of this study was to compare the value of static preload indexes (echocardiographic measures of RV and LV size) with the value of an echocardiographic marker of SV variation (dynamic preload response index) under conditions of progressive blood loss and restitution in rabbits receiving mechanical ventilation.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Animals. Twelve New Zealand female rabbits (Charles River Laboratories ESD; Romans, France) weighing 3-4.5 kg were used in this study. They were housed for at least 1 wk before study in separate cages under controlled conditions (21 ± 1°C; 12:12-h light-dark cycle) and received standard rabbit chow diet and tap water ad libitum. Initial anesthesia consisted of intravenous midazolam (1 mg/kg) and ketamine (0.75 mg/kg) into an ear vein through a Teflon catheter. Adequate anaesthesia was confirmed by absence of eyelid reflex, changes in heart rate and arterial pressure, or movement after a tail clamp application. After the dissection of the trachea, a tracheostomy was performed (horizontal section at the level of the second or third tracheal annulus), a pediatric endotracheal tube was connected to a ventilator, and pancuronium bromide (8 mg) was injected. A continuous infusion of midazolam (1 mg · kg¹ · h¹) maintained anesthesia; 0.9% NaCl (6 ml · kg¹ · h¹) and intermittent injections of pancuronium were administered throughout the study. Mechanical ventilation was initiated, and the tidal volume was adjusted to obtain a peak respiratory pressure of 20 cmH₂O. The ventilation rate and fraction of inspired oxygen were fixed, respectively, at 40 cycles/min and 21%, and further changes in tidal volume were not allowed during the entire protocol. A catheter filled with normal saline was inserted into the right femoral vein. Animal care conformed to the Helsinki Declaration, and the study was conducted according to the regulations of the French Ministry of Agriculture.

Experimental protocol. The rabbit was placed in a supine position, and mild hypovolemia was induced by a stepwise cumulative withdrawal of 5, 10, 15, 20, 25, and 30 ml of blood through the catheter. Blood was withdrawn into a sterile heparinized (5,000 U/l) blood bag and maintained at 37°C by a heated blanket. The time interval between each step was 10 min. All of the withdrawn blood was reinjected 10 min after the last bleeding step.

Echocardiographic measurements. Commercially available echocardiographic equipment (Sonos 5500, Hewlett-Packard; Les Ullis, France) with a 8- to 12-MHz pediatric transthoracic transducer was used for all the studies. All M-mode, two-dimensional, and Doppler measurements were recorded at each step after 10 min of stabilization, on paper at a speed of 100 mm/s, and on 0.5-in. SVHS videotape. All measurements were made on-line with the Sonos 5500 software system. In the supine position, from the parasternal long-axis view, M-mode images of the LV and RV chambers were obtained at the papillary muscle level. LV end-diastolic diameter (LVEDD; maximal diameter), LV systolic diameter (LVSD, at the maximal systolic movement of the septum), and RV end-diastolic diameter (RVEDD) were measured at the end of the expiration. From the two-dimensional view, we measured aortic diameter (D) during inspiration and expiration at the level of the aortic annulus at baseline and at each blood withdrawal step. From the apical five-chamber view, pulsed Doppler aortic flow was recorded at the level of the annulus, and the aortic velocity time integral (VTI_Ao) was measured. Each echocardiographic and Doppler index measurement is presented as the average of three measurements. SV was determined as follows: SV = VTI_Ao × (D²)/4. Maximal and minimal values of VTI_Ao (VTI_Aomax and VTI_Aomin, respectively) were determined over a single respiratory cycle. Respiratory changes in VTI_Ao (%VTI_Ao) were calculated as follows: %VTI_Ao = (VTI_Aomax  VTI_Aomin)/average of VTI_Aomax and VTI_Aomin. These indexes were measured in triplicate over three consecutive respiratory cycles and averaged. In six rabbits, two sets of all measurements were recorded to analyze the intraobserver reproducibility.

Statistical analysis. One-way ANOVA for repeated measures was performed to compare the different steps. Individual withdrawal curves were studied, and linear regressions between each variable and blood withdrawal volume were calculated to analyze the accuracy of echocardiographic parameters for assessing blood loss. Coefficients of regression (r) were averaged. Individual and normalized slopes (divided by the maximal value) were recorded and averaged to determine the sensitivity of each echocardiographic parameter. A whole group analysis was done, and a correlation between each parameter at 30 ml and changes in AV after the blood restitution were assessed. A correlation was assessed between blood withdrawal and respiratory changes of VTI_Ao using all the measurements. A P value of <0.05 was considered statistically significant. All values are presented as means ± SD. Reproducibility of measurement was assessed as bias ± limits of agreement as described by Bland and Altman (3).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Echocardiographic measurements. Heart rate remained unchanged throughout the study (from 261 ± 33 beats/min at the baseline to 251 ± 35 beats/min after 30 ml of blood withdrawal). VTI_Ao progressively decreased from 6.5 ± 0.9 cm at baseline to 4.4 ± 1.2 cm (P < 0.0001) after the withdrawal of 30 ml of blood. SV and cardiac output (CO) also decreased (from 2.13 ± 0.6 to 1.46 ± 0.54 ml, P < 0.02, and from 0.56 ± 0.16 to 0.37 ± 0.16 l/min, P < 0.02) and returned to the prewithdrawal levels after blood restitution (Fig. 1). Changes in these parameters became statistically significant after 10 ml of blood withdrawal for VTI_Ao and after 25 ml for SV and CO. Respiratory variations in VTI increased from 6 ± 5% at the baseline to 36 ± 10% (P < 0.0001) after a blood withdrawal of 30 ml and returned to baseline value after blood restitution (Fig. 1). These differences were statistically significant at the first withdrawal step.

View larger version (20K): [in this window] [in a new window]   Fig. 1.   Stroke volume (SV) and respiratory changes of the aortic velocity time interval (%VTIAo) at different steps of blood withdrawal. B, baseline. *P < 0.05 vs. basal conditions.

View larger version (21K): [in this window] [in a new window]   Fig. 2.   Left (LV) and right ventricular (RV) diastolic diameters at different steps of blood withdrawal. LVEDD, LV end-diastolic diameter; RVEDD, RV end-diastolic diameter. *P < 0.05 vs. basal conditions.

Correlations. Table 1 presents the correlation coefficients between echocardiographic parameters and the volume of blood withdrawal. A good correlation was demonstrated between the amount of blood withdrawal and the %VTI_Ao (r = 0.8, P < 0.001). A strong correlation (Fig. 3) was found between %VTI_Ao at the 30-ml step of blood withdrawal and the percent increase in SV after blood restitution (r = 0.83, P < 0.0001). The other indexes obtained at the last withdrawal step were not correlated with changes in CO after blood restitution.

View larger version (13K): [in this window] [in a new window]   Fig. 3.   Relationship between %VTIAo at the last withdrawal step (30 ml) and changes in SV after retransfusion.

Slopes. The slopes and the normalized slopes of the relationships between the volume of blood withdrawal and echocardiographic parameters are presented in Table 1. All slopes were ~1% except for %VTI_Ao, which was 16.6 ± 5.6%.

Reproducibility. When expressed as bias ± precision, the reproducibility of echocardiographic parameters was excellent: LVEDD = 0.06 ± 0.19 cm; RVEDD = 0.04 ± 0.1 cm; and VTI = 0.11 ± 0.66 cm.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In this study, the dynamic and static echocardiographic indexes of ventricular preload were compared during a stepwise hypovolemia and after reinfusion. We demonstrated, during hypovolemia, a progressive reduction in ventricular preload indexes as well as an increased magnitude of %VTI_Ao. Moreover, we learned whether or not these indexes could predict the hemodynamic responses to a loading challenge induction (retransfusion). However, during the sequential blood reinfusion, only the %VTI_Ao correlated with the increase in systemic blood flow, suggesting that dynamic parameter of preload responsiveness is more reliable than static parameters for assessing the hemodynamic response to volume infusion. Several studies have demonstrated that changes in preload produce proportional LVEDD changes (6, 17, 9). Dalibon et al. (6) studied the effect of a graded hemorrhage on LV end-diastolic area in pigs using transthoracic echocardiography. They demonstrated that this index was very sensitive to acute blood volume losses. In 35 anesthetized patients, Cheung et al. (4) showed that the transesophageal LV end-diastolic area decreased during graded hypovolemia. A mean change of 0.3 cm²/1% of estimated blood volume was observed in patients with or without wall motion abnormalities (4). Reich et al. (14) demonstrated that a 5-8% decrease in blood volume significantly reduced the LV end-diastolic area. The results of this present study are consistent with all of these findings because LVEDD was found to be highly sensitive to blood withdrawal. On the other hand, other clinical studies have shown that volume infusion might result in increased LV dimensions (5, 19). In a study of surgical patients, preoperative hypervolemic hemodilution resulted in an increased LV end-diastolic area using transesophageal echocardiography (19). Similar results have been reported in patients after aortic surgery receiving a colloid infusion of 250 ml (5). However, divergent results were reported by Axler et al. (1). With the use of transesophageal echocardiography in critically ill patients, the latter group showed that a rapid infusion of 250-500 ml of saline did not significantly increase LV end-diastolic area (1). These authors speculated that a variation of ~10% of SV due to an increased LV volume is associated with a LVEDD change of <1 mm. This variation is below the limit of the reproducibility of this technique. Moreover, changes in LVEDD are dependent on LV stiffness. In the presence of reduced cardiac compliance, the LV filling pressure-diameter relationship is steep, and large variations of preload increase the LV pressure, leading to a small variation of LVEDD. Thus the use of LV size to predict a hemodynamic effect of fluid loading may be a subject of debate. Swenson et al. (15) demonstrated a significant correlation when comparing LV end-diastolic area with changes in CO. Others (8, 16, 18) examined mechanically ventilated patients to determine whether LV end-diastolic area, measured by transesophageal echocardiography, could identify those ventilated patients who increased SV by 15-20% or more (i.e., responders) after volume expansion. Although a smaller initial value of LV end-diastolic area occurred in responders (as compared with nonresponders) in two studies (16, 18), neither of these two studies demonstrated the existence of a threshold LV end-diastolic area value below which a large proportion of ventilated patients responded to volume administration (8, 16, 18). One of the main reasons explaining these findings is that the preload/SV relationship depends mainly on the value of systolic function. Thus, in a patient with reduced systolic function, an increase in preload induced by fluid infusion will result in a less marked increase in SV (flat portion of the curve) compared with a patient with normal systolic function (steep portion of the curve). In our study, no close relationship was observed between the value of LVEDD at the last step of hemorrhage and the percent increase of CO after blood restitution. Therefore, LVEDD seems to be a poor indicator of fluid responsiveness.

In the present study, RV end-diastolic size decreased with blood withdrawal. Diebel et al. (7) also demonstrated a lower RV end-diastolic volume as assessed by a modified pulmonary artery catheter. However, a large overlap between individual baseline values of RV end-diastolic volume was observed among their responders and nonresponders. In the present study, using a noninvasive technique, the value of RVEDD before blood restitution was not related to the effect of blood reinfusion on CO.

We found that %VTI_Ao was more sensitive to graded hemorrhage than any of the other echocardiographically derived indexes. Moreover, this parameter was the only one found to be a good predictor of hemodynamic response after retransfusion. Thus the magnitude of the respiratory changes in LV SV can be a marker of fluid responsiveness. By decreasing the venous return pressure gradient, mechanical insufflation may decrease the RV SV if the RV is sensitive to changes in preload. Under this condition, the decreased LV filling may result in a decrease in LV SV. Therefore, the magnitude of the respiratory changes in LV SV should reflect the sensitivity of the heart to changes in preload induced by mechanical insufflation. In favor of this concept, the respiratory changes in arterial systolic or pulse pressure that reflect the respiratory changes in LV SV have been shown to accurately predict fluid responsiveness in mechanically ventilated patients (11, 13). In contrast, both pulmonary artery occlusion pressure and right atrial pressure at baseline failed to predict fluid responsiveness (11). In another study involving septic patients receiving mechanical ventilation, the magnitude of the respiratory changes in systolic pressure has been demonstrated to assess fluid responsiveness better than the baseline value of LV end-diastolic area measured with transesophageal echocardiography (16). Feissel et al. (8) demonstrated that the magnitude of respiratory variations of aortic blood velocity was a more accurate method than the measurement of LV dimensions for predicting the hemodynamic effects of volume expansion in septic shock patients under mechanical ventilation. In that study, aortic blood velocity was measured using transesophageal echocardiography at the level of the aortic annulus. In our study, using transthoracic echocardiography, we directly determined respiratory changes in SV through respiratory changes in VTI_Ao. Indeed, because the SV was calculated using VTI_Ao and because annulus diameter remained unchanged over the respiratory cycle, we assumed that respiratory changes in VTI reflected respiratory changes in SV.

The present study may involve certain limitations. First, although measurements of CO by cardiac Doppler have been well validated in patients, few data are available in small animals. Recently, a good agreement between cardiac Doppler and ultrasound transit time measurements of CO was reported (2). Second, anesthesia have maybe depressed sympathetic activity as reflected by the absence of tachycardia during the progressive hemorrhage. Therefore, our data cannot be applied readily to less anaesthetized or awake animals. Finally, this study involved normal rabbit hearts, and our findings should not be extrapolated at this time to patients or animals with cardiac abnormalities.

In conclusion, in this mechanically ventilated rabbit model, the stepwise blood withdrawal period resulted in a decrease in RV and LV dimensions and CO, whereas the magnitude of the respiratory variations in VTI_Ao increased. Although all echocardiographic parameters returned to baseline values after blood restitution, only the magnitude of the respiratory variations in VTI_Ao accurately predicted the percent increase in CO in response to blood infusion. These findings underline the superiority of dynamic indexes of preload responsiveness over static markers of cardiac preload measured using echocardiography in the detection of hypovolemia and fluid responsiveness in subjects receiving mechanical ventilation.