Fluid responsiveness in mechanically ventilated patients: a review of indices used in intensive care

Karim Bendjelid
Jacques-A. Romand
Introduction
Hypotension is one of the most frequent clinical signs observed in critically ill patients. To restore normal blood pressure, the cardiovascular filling (preloaddefined as end-diastolic volume of both ventricular chambers), cardiac function (inotropism), and vascular resistance (afterload) must be assessed. Hemodynamic instability secondary to effective or relative intravascular volume depletion are very common, and intravascular fluid resuscitation or volume expansion (VE) allows restoration of ventricular filling, cardiac output and ultimately arterial blood pressure [1, 2]. However, in the Frank-Starling curve (stroke volume as a function of preload) the slope presents on its early phase a steep portion which is followed by a plateau (Fig. 1). As a consequence, when the plateau is reached, vigorous fluid resuscitation carries out the risk of generating volume overload and pulmonary edema and/or right-ventricular dysfunction. Thus in hypotensive patients methods able to unmask decreased preload and to predict whether cardiac output will increase or not with VE have been sought after for many years. Presently, as few methods are able to assess ventricular volumes continuously and directly, static pressure measurements and echocardiographically measured ventricularend-diastolic areas are used as tools to monitor cardiovascular filling. Replacing static measurements, dynamic monitoring consisting in assessment of fluid responsiveness using changes in systolic arterial pressure, and pulse pressure induced by positive pressure ventilation have been proposed. The present review analyses the current roles and limitations of the most frequently used methods in clinical practice to predict fluid responsiveness in patients undergoing mechanical ventilation (MV) (Table 1). One method routinely used to evaluate intravascular volume in hypotensive patients uses hemodynamic response to a fluid challenge [3]. This method consists in infusing a defined amount of fluid over a brief period of time. The response to the intravascular volume loading can be monitored clinically by heart rate, blood pressure, pulse pressure (systolic minus diastolic blood pressure), and urine output or by invasive monitoring with the measurements of the right atrial pressure (RAP), pulmonary artery occlusion pressure (Ppao), and cardiac output. Such a fluid management protocol assumes that the intravascular volume of the critically ill patients can be defined by the relationship between preload and cardiac output, and that changing preload with volume infusion affects cardiac output. Thus an increase in cardiac output following VE (patient responder) unmasks an hypovolemic state or preload dependency. On the other hand, lack of change or a decrease in cardiac output following VE (nonresponding patient) is attributed to a normovolemic, to an overloaded, or to cardiac failure state. Therefore, as the fluid responsiveness defines the response of cardiac output to volume challenge, indices which can predict the latter are necessary. Static measurements for preload assessment
Measures of intracardiac pressures
According to the Frank-Starling law, left-ventricular preload is defined as the myocardial fiber length at the end Table 1 Studies of indices used as bedside indicators of preload reserve and fluid responsiveness in hypotensive patients under positive-pressure ventilation (BMI body mass index, CO cardiac output, CI cardiac index, SV stroke volume, SVI stroke volume index, IAC invasive arterial catheter, MV proportion of patients mechanically ventilated, . increase, . decrease, PAC pulmonary artery catheter, R responders, NR nonresponders, FC fluid challenge, HES hydroxyethyl starch, RL Ringer’s lactate, Alb albumin, .down delta down, .PP respiratory variation in pulse pressure, LVEDV left-ventricular end diastolic volume, SPV systolic pressure
variation, SVV stroke volume variation, TEE transesophageal echocardiography, Ppao pulmonary artery occlusion pressure, RAP right atrial pressure, RVEDV right-ventricular-end diastolic volume, FC fluid challenge) Variable Tech- n MV Volume (ml) Duration Definition Definition p: Refermeasured nique (%) and type of of FC of R of NR difference ence plasma substitute (min) in baseline values R vs. NR
Rap PAC 28 46 250 Alb 5% 20–30 . SVI . SVI or unchanged NS 37
Rap PAC 41 76 300 Alb 4.5% 30 . CI CI . or unchanged NS 18
Rap PAC 25 94.4 NaCl 9‰ + Until .Ppao . SV =10% . SV <10% 0.04 31
Alb 5% to . Ppao
Rap PAC 40 100 500 HES 6% 30 . CI >15% . CI <15% NS 36
Ppao PAC 28 46 250 Alb 5% 20–30 . SVI . SVI or unchanged NS 37
Ppao PAC 41 76 300 Alb 4.5% 30 . CI CI . or unchanged NS 18
Ppao PAC 29 69 300–500 RL ? bolus . C0>10% C0 . or unchanged <0.01 40
Ppao PAC 32 84 300–500 RL ? . CI >20% . CI <20% NS 41
Ppao PAC 16 100 500 HES 6% 30 . CI >15% . CI <15% 0.1 42
Ppao PAC 41 100 500 pPentastarch 15 . SV =20% . SV <20% 0.003 25
Ppao PAC 25 94.4 NaCl 9‰, Until .Ppao . SV =10% . SV <10% 0.001 31
Alb 5% to. Ppao
Ppao PAC 40 100 500 HES 6% 30 . CI >15% . CI <15% NS 36
Ppao PAC 19 100 500–750 HES 6% 10 . C0>10% . SV <10% 0.0085 39
RVEDV PAC 29 69 300–500 RL ? bolus . C0>10% C0 . or unchanged <0.001 40
RVEDV PAC 32 84 300–500 RL ? . CI >20% . CI <20% <0.002 41
RVEDV PAC 25 94.4 NaCl 9‰, Until .Ppao . SV =10% . SV <10% 0.22 31
Alb 5% to. Ppao
LVEDV TEE 16 100 500 HES 6% 30 . CI >15% . CI <15% 0.005 42
LVEDV TEE 41 100 500 Pentastarch 15 . SV =20% . SV <20% 0.012 25
LVEDV TEE 19 100 8 ml/kg HES 6% 30 . CI >15% . CI <15% NS 79
LVEDV TEE 19 100 500–750 HES 6% 10 . C0>10% . SV <10% NS 39
SPV IAC 16 100 500 HES 6% 30 . CI >15% . CI <15% 0.0001 42
SPV IAC 40 100 500 HES 6% 30 . CI >15% . CI <15% <0.001 36
SPV IAC 19 100 500–750 HES 6% 10 . C0>10% . SV <10% 0.017 39
.down IAC 16 100 500 HES 6% 30 . CI >15% . CI <15% 0.0001 42
.down IAC 19 100 500–750 HES 6% 10 . C0>10% . SV <10% 0.025 39
.PP IAC 40 100 500 HES 6% 30 . CI >15% . CI <15% <0.001 36
Fig. 1 Representation of Frank-Starling curve with relationship between
ventricular preload and ventricular stroke volume in patient X. After volume expansion the same magnitude of change in preload recruit less stroke volume, because the plateau of the curve is reached which characterize a condition of preload independency of the diastole. In clinical practice, the left-ventricular end-diastolic volume is used as a surrogate to define leftventricular preload [4]. However, this volumetric parameter is not easily assessed in critically ill patients. In normal conditions, a fairly good correlation exists between ventricular end-diastolic volumes and mean atrial pressures, and ventricular preloads are approximated by RAP and/or Ppao in patients breathing spontaneously [5, 6]. Critically ill patients often require positive pressure ventilation, which modifies the pressure regimen in the thorax in comparison to spontaneous breathing. Indeed, during MV RAP and Ppao rise secondary to an increase in
intrathoracic pressure which rises pericardial pressure. This pressure increase induces a decrease in venous return [7, 8] with first a decrease in right and few heart beats later in left-ventricular end-diastolic volumes, respectively [9, 10]. Under extreme conditions such as acute severe pulmonary emboli and/or marked hyperinflation, RAP may also rise secondary to an increase afterload of the right ventricle. Moreover, under positive pressure ventilation not only ventricular but also thoracopulmonary compliances and abdominal pressure variations are observed over time. Thus a variable relationship between cardiac pressures and cardiac volumes is often observed [11, 12, 13, 14]. It has also been demonstrated that changes in intracardiac pressure (RAP, Ppao) no longer directly reflect changes in intravascular volume [15]. Pinsky et al. [16, 17] have demonstrated that changes in RAP do not follow changes in right-ventricular
end-diastolic volume in postoperative cardiac surgery patients under positive pressure ventilation. Reuse et al. [18] observed no correlation between RAP and right-ventricular end-diastolic volume calculated from a thermodilution technique in hypovolemic patients before and after fluid resuscitation. The discordance between RAP and right-ventricular end-diastolic volume measurements may result from asystematic underestimation
of the effect of positive-pressure ventilation on the right heart [16, 17]. Nevertheless, the RAP value measured either with a central venous catheter or a pulmonary artery catheter is still used to estimate preload and to guide intravascular volume therapy in patient under positive pressure ventilation [19, 20]. On the left side, the MV-induced intrathoracic pressure changes, compared to spontaneously breathing, only minimally alters the relationship between left atrial pressure and left-ventricular end-diastolic volume measurement in postoperative cardiac surgery patients [21]. However, several other studies show no relationship between Ppao and left-ventricular end-diastolic volume measured by either radionuclide angiography [12, 22], transthoracic echocardiography (TTE) [23], or transesophageal echocardiography (TEE) [24, 25, 26]. The latter findings may be related to the indirect pulmonary artery catheter method for assessing left atrial pressure [27, 28], although several studies have demonstrated that Ppao using PAC is a reliable indirect measurement of left atrial pressure [29, 30] in positive-pressure MV patients. Right atrial pressure used to predict fluid responsiveness Wagner et al. [31] reported that RAP was significantly lower before volume challenge in responders than in nonresponders (p=0.04) when patients were under positive pressure ventilation. Jellinek et al. [32] found that a RAP lower than 10 mmHg predicts a decrease in cardiac index higher than 20% when a transient 30 cm H2O increase in intrathoracic pressure is administrated. Presuming that the principle cause of decrease in cardiac output in the latter study was due to a reduction in venous return [9, 33, 34, 35], RAP predicts reverse VE hemodynamic effect. Nevertheless, some clinical investigations studying fluid responsiveness in MV patients have reported that RAP poorly predicts increased cardiac output after volume expansion [18, 36, 37]. Indeed, in these studies RAP did not differentiate patients whose cardiac output did or did not increase after VE (responders and nonresponders, respectively)