Ultrasonographic caval indices do not significantly contribute to predicting fluid responsiveness immediately after coronary artery bypass grafting when compared to passive leg raising

Postoperative goal-directed therapy has been shown to substantially improve clinical outcomes in surgical patients [1, 2]. Two important aspects of a goal-directed protocol include definition of hemodynamic target and early initiation of the therapeutic measures [3, 4]. Appropriate fluid management is one of the most important elements of early goal-directed therapy after cardiothoracic surgery [5–7]. Therefore, the fundamental issue in a goal-directed volume therapy remains determination of the potential response to fluid challenge, e.g. fluid responsiveness [8, 9].

Among the numerous monitoring methods, ultrasonographic measurement of inferior vena cava (IVC) diameter respiratory variation seems to meet the criteria of an ideal bedside tool to assess the individual fluid responsiveness. However, this method has been validated in selected populations (e.g. hemodialysis, septic shock, subarachnoid hemorrhage) or healthy volunteers, and its usefulness has not been confirmed in patients after cardiac surgery [10–15]. In our recent study, IVC-derived dynamic indices failed to predict fluid responsiveness in the first six hours after cardiothoracic procedures [16]. Nevertheless, heterogeneity of the study population (different cardiac surgery procedures, concomitant valve disease) have accounted to the study results. Usefulness of ultrasonographic measurement of IVC respiratory variation in selected cardiosurgical patients (e.g. coronary artery bypass grafting, preserved left ventricular systolic function) remains still an open question.

The purpose of the present study was to assess the usefulness of dynamic IVC-derived parameters (collapsibility index, distensibility index) in comparison to passive leg raising, in postoperative fluid management in mechanically ventilated patients with left ventricular ejection fraction ≥30 %, immediately after elective coronary artery bypass grafting.

This prospective observational case series was designed to test the effectiveness of IVC-derived parameters in predicting fluid responsiveness immediately after elective coronary artery bypass grafting. Our previous study did not confirm the utility of caval indices as a guiding tool for early postoperative fluid therapy in a heterogeneous group after cardiac surgery [16]. The present study was limited to patients qualified for elective coronary artery bypass surgery. To reduce any potential bias, severely diminished left ventricular ejection fraction (LVEF <30 %), LV dilatation, at least moderate aortic or mitral valve disease, impaired right ventricular systolic function (TAPSE <16 mm) and severe tricuspid regurgitation were the exclusion criteria from the study. All patients were mechanically ventilated with standard parameters to limit influence of spontaneous breathing on IVC diameter. The study protocol assumed cardiac output response (increase of ≥15 % compared to the baseline) after intravenous administration of 250 ml of saline as a reference to assess fluid responsiveness. Additionally, passive leg raising was used as an alternative reversible volume challenge in all patients.

The majority of patients in our study (almost 70 %) responded to fluid challenge with significant cardiac output increase and met the criteria of fluid responsiveness. While we excluded patients with severely impaired left ventricular systolic function and LV dilatation, the main contributor of stroke volume increase were preload conditions. It seems that response to fluid challenge was mainly determined by absolute (preoperative fluid restriction, intraoperative blood loss) and relative hypovolemia (increased vascular bed capacity induced by hypothermia, vasoactive agents and anesthetic drugs, as well as fluid shifts). In our study fluid challenge with 250 ml of saline/PLR was safe and well tolerated in all subjects. Considering our results, we assume that immediate administration of crystalloids directly after CABG regardless hemodynamic parameters is reasonable, low-risk profile management.

Ultrasonographic measurement of IVC respiratory variation did not reach statistical significance in differentiation between fluid responders and nonresponders. Although we observed a trend towards higher values of both IVC indices (IVC-CI and IVC-DI) in responder group, none of them was statistically significant. In our study material we observed high variation of recorded measurements with overlapping series of values. Thus, it was impossible to determine a cut-off point separating responders and nonresponders.

Ultrasonographic assessment of the inferior vena cava diameter has been proposed as a bedside, noninvasive marker of fluid status [10]. The usefulness of this method has been validated during hemodialysis and continuous ultrafiltration or in patients with septic shock [11–14]. IVC distensibility index has been proved to predict fluid responsiveness in mechanically ventilated patients with septic shock and subarachnoid hemorrhage (with cut-off value of 18 %) [10, 15]. However, conflicting results of several prospective studies in unselected patients indicated the need for a cautionary use of IVC ultrasound in everyday ICU/emergency practice. Muller et al. [18] revealed that in spontaneously breathing patients with acute circulatory failure, high IVC-CI values (>40 %) were usually associated with fluid responsiveness, however values <40 % did not exclude positive response to volume expansion. Corl et al. [19] did not confirm utility of IVC-CI in predicting fluid responsiveness in their heterogeneous sample of adult emergency department patients. Juhl-Olsen et al. showed no correlation between IVC-CI and magnitude of hemodynamic response to early hemorrhage in healthy blood donors [20].

There are multiple possible reasons contributing to limited usefulness of dynamic IVC derivatives in early postoperative period after CABG [21]. First of all, there are numerous factors influencing IVC diameter, its respiratory variation and vascular bed capacity, such as cardiopulmonary bypass, intraoperative hypothermia, anesthetics and vasoactive agents. Also surgical chest opening and postoperative suction drainage (with negative pressure of 15–20 cmH₂O) may contribute to negative results of this paper. However very attractive, ultrasonographic definition of fluid responsiveness using IVC respiratory variation failed in early postoperative period after cardiothoracic surgery that was also described in our previous work [16].

Several diagnostic approaches have been tested to determine fluid responsiveness in mechanically ventilated patients. Respiration-induced changes in arterial pulse pressure (ΔPP), based on the variation in arterial pressure associated with mechanical ventilation, have been demonstrated to accurately predict fluid responsiveness in mechanically ventilated patients with regular heart rhythm, who are making no respiratory efforts [22]. However, ΔPP is of poor value to predict fluid responsiveness in patients triggering the ventilator or in the presence of arrhythmias [23]. Additionally, reliable calculation of ΔPP is time-consuming, requires experience and specialized computer software. Therefore, this method is not commonly used for navigating the fluid therapy in the intensive care unit. Several publications have proposed using PLR maneuver to predict preload responsiveness [24]. Passive leg raising transfers about 300 ml of blood from the lower limbs to the intrathoracic compartment, mimicking fluid expansion [24–26]. PLR maneuver is quick, easy to perform at the bedside and completely reversible (with no side effects connected with actual fluid challenge). Proper PLR performance is essential for its reliability. Monnet and Teboul recently described 5-step protocol of PLR [25]. The most important issues are: starting the test from the semirecumbent position (adding trunk lowering to leg raising mobilizes venous blood from the splanchnic compartment and magnifies the effects on cardiac preload) and direct measurement of cardiac output at all steps. In our study, PLR was an equivalent of intravenous administration of 250 ml of saline. Change in cardiac output after passive leg raising correlated well with that induced by volume expansion (p = 0.000, r = 0.822). A 15 % PLR-induced increase of cardiac output predicted the corresponding increase in cardiac output after fluid administration with 79.17 % sensitivity and 81.82 % specificity. These results are in concordance with available literature. Monnet et al. [26] proved that transient hemodynamic changes induced by PLR had an excellent prediction in preload responsiveness in mechanically ventilated ICU patients. In that study, a > 10 % increase of aortic flow induced by PLR was predictive of an increase of aortic flow of >15 % in response to volume expansion. Lamia et al. [27] demonstrated that a PLR-induced increase in stroke volume of ≥12.5 % predicted an increase in stroke volume of ≥15 % after fluid challenge. In the study of Maizel et al. [24], PLR-induced increase of stroke volume/cardiac output by >12 % was highly predictive of volume responsiveness in spontaneously breathing patients. The meta-analysis of 21 studies conducted in 991 adult patients, published recently by Monnet et al. [28], also demonstrated that the changes in CO during a PLR test predicted fluid responsiveness with excellent pooled sensitivity and specificity. The pooled area under the ROC curve in the meta-analysis was 0.95 ± 0.01 [28]. The AUROC in our study (0.805) confirmed very good diagnostic accuracy of the test.

Reliable, real time measurement of stroke volume (implying accurate calculation of cardiac output) remains the key issue in determination of fluid responsiveness [29]. The thermodilution technique has been validated in a variety of experimental models and thermodilution pulmonary artery catheters (PACs) were adopted as the clinical reference standard and subsequently used in the comparison studies [29]. However, the thermodilution devices are invasive, slow-reacting, require stable core temperatures and cannot be used in patients with tricuspid insufficiency. Additionally, there has been observed a decline in PACs use due to the results of prospective, randomized controlled trials assessing the efficacy of PACs in critically ill patients [30–32]. Their results did not justify the widespread use of thermodilution devices in ICU. There are numerous minimally or noninvasive devices available to measure cardiac output in the ICU setting [29]. Bioimpedance, pulse contour analysis, partial rebreathing and pulse wave velocity analysis require technical support and operator experience [29]. Doppler-based echocardiographic techniques are less invasive (transesophageal) or noninvasive (transthoracic), offer real-time beat-to-beat measurements and have similar accuracy as thermodilution pulmonary artery catheters [17, 29]. Despite some technical issues (operator experience, learning curve, accurate ultrasound beam alignment), Doppler echocardiography seems an ideal tool for bedside cardiac output assessment.

We conclude that dynamic IVC-derived caval index does not predict fluid responsiveness in mechanically ventilated patients with preserved LVEF immediately after elective coronary artery bypass grafting. Passive leg raising is not inferior to intravenous volume expansion in differentiating between fluid responders and nonresponders. Immediate fluid challenge after CABG is safe and well tolerated.

AUROC, area under the receiver operator characteristics curve; CABG, coronary artery bypass grafting; CVP, central venous pressure; DBP, diastolic blood pressure; HR, heart rate; ICU, intensive care unit; IVC-CI, inferior vena cava collapsibility index; IVC-DI, inferior vena cava distensibility index; IVC, inferior vena cava; LVEF, left ventricular ejection fraction; LVOT, left ventricular outflow tract; MAP, mean blood pressure; PAC, pulmonary artery catheter; PLR, passive leg raising, PP-pulse pressure; SBP, systolic blood pressure; SD, standard deviation; VTI, velocity-time integral