Volume Responsiveness: What It Does Not Tell Us

IN THIS ISSUE of the Journal of Cardiothoracic and Vascular Anesthesia, Zaky et al. present an article entitled “End-of-Procedure Volume Responsiveness Defined by the Passive Leg Raise Test Is Not Associated With Acute Kidney Injury After Cardiopulmonary Bypass.” In this study the investigators tested for volume responsiveness by performing passive leg raising (PLR) on 129 cardiac surgical patients at the conclusion of the procedure, prior to transport to the intensive care unit. Their hypothesis was that volume responsiveness at the conclusion of cardiac surgery is predictive of perioperative acute kidney injury (AKI). An implied hypothesis is that volume responsiveness is associated with hypovolemia, which, in turn, is associated with poor renal perfusion and AKI. Their study found the overall incidence of AKI to be 30%, and there was no difference detected in the incidence of AKI in volume responders versus nonresponders. Although their negative results are important, they are not surprising.

The term “volume responsive” refers to a positive effect on hemodynamics (usually cardiac stroke volume) resulting from administration of fluid. There have been numerous definitions proposed, with one of the more popular being an increase in cardiac output or stroke volume by 10% to 15% following a 500-mL fluid bolus to an adult. This metric can be useful in solving hemodynamic problems involving low cardiac output and has been central to numerous goal-directed fluid therapy (GDFT) algorithms. Because administration of fluid can come at a clinical price (eg, fluid overload, cardiac failure), the administration of fluid simply as a test of fluid responsiveness is suboptimal. Far preferable is to predict fluid responsiveness using other means, such as dynamic parameters (eg, stroke-volume variation, pulse-pressure variation) or passive leg raising (PLR). The use of dynamic parameters to predict fluid responsiveness in cardiac surgery may be limited by the presence of cardiac arrhythmias and the open chest, and PLR usually is not possible during cardiac surgery. However, observation of hemodynamic changes in response to Trendelenburg position may be used instead.

If volume responsiveness is associated with hypovolemia and poor renal perfusion, then why did Zaky et al. not find a higher incidence of AKI in the volume-responsive group? One answer to this question is that volume responsiveness is a nonspecific indicator of intravascular volume and tissue perfusion; ie, many subjects who are volume responders are not hypovolemic and are in good hemodynamic condition. Likewise, many patients who are not volume responsive are euvolemic. Volume responsiveness does not indicate anything other than that administration of fluid would increase cardiac stroke volume. As important as it is to understand the potential use of volume responsiveness in clinical management, it is equally important to appreciate its limitations. There are many possible reasons for the development of postoperative AKI, with some of the more common being decreased cardiac output either from hypovolemia or cardiac failure, low perfusion pressure from vasodilatation or vasoplegia, intense arterial vasoconstriction, and emboli. Importantly, decreased cardiac output from hypovolemia is the only one of these that consistently exhibits volume responsiveness.

The key to applying volume responsiveness to clinical care is to appreciate the context in which it is present. Certainly if a patient exhibits low cardiac output with normal cardiac function and is volume responsive as assessed by PLR, administration of fluid is appropriate. On the other hand, most healthy subjects with normal cardiac output and renal perfusion are volume responsive as well (it is the rare healthy person whose left ventricle performs at the top of its Starling curve). This consistently has been demonstrated in volume-responsiveness studies of healthy volunteers. Therefore, volume responsiveness does not define hypovolemia, and it does not provide information about adequacy of renal perfusion. Other measurements, including cardiac output, lactate, blood pH, urine output, and mixed venous oxygenation provide more direct information about organ perfusion and oxygenation. Zaky et al. point out that assessment of the renal microcirculation and sensitive biomarkers will aid in understanding perturbations in renal perfusion in the future. The mere fact that a patient potentially is volume responsive does not mean he or she needs or should receive fluid.

Volume responsiveness has been used extensively in GDFT protocols, with some protocols recommending administration of fluid whenever volume responsiveness is present. In these protocols (lack of) volume responsiveness is used not as a diagnostic tool, but as a goal or endpoint. Rather than use such metrics as cardiac output or oxygen delivery, the process in these protocols is to simply administer fluid until the patient is no longer volume responsive. A recent study, using the dynamic parameter “pleth variability index” as a primary criterion for fluid administration in orthopedic surgery, revealed no clinical benefit of this approach, with study patients receiving more perioperative fluid than controls. A study of colonic surgery patients by Challand et al. used an algorithm calling for fluid administration if a 200-mL bolus any time a previous bolus resulted in a >10% increase in SV. The GDFT group received significantly more fluid and had a longer hospital length of stay than controls, and the effect was particularly pronounced in the cardiovascularly fit group. This study illustrated that most fit patients will increase their stroke volume when given a fluid bolus, and if such boluses are given repeatedly, without regard to any other clinical indicators, volume overload will result. In cardiac surgery, as well as general surgery, relative fluid restriction is advocated, and if a goal-directed approach is used, the goals should be hemodynamic stability and adequate oxygen delivery to the vital organs and tissues. A recent meta-analysis indicated that use of fluid responsiveness metrics alone is insufficient to achieve improved outcome from GDFT. These authors suggest ed the use of additional indices, such as cardiac output and oxygen delivery. The use of advanced indices, such as stroke volume and oxygen delivery as goals in GDFT, has shown improved outcome in high-risk cardiac surgical patients.

As Zaky et al. point out, excess fluid administration resulting from GDFT implementation can increase the incidence of AKI.^, This author believes the adage “Give enough fluid, but not a drop more” applies, because fluid overload can lead to heart failure in the cardiac surgical population, as well as AKI, increased lung water, and edema affecting the airway. Certainly, then, providing fluid boluses simply because volume responsiveness is present is ill advised, particularly in cardiac surgical patients in whom cardiac and other organ failures are such high risks. This leads to the conclusion that volume responsiveness should not be treated as an isolated, independent metric, and its presence should not always prompt fluid administration.

Volume responsiveness can be very useful for clinical problem solving, particularly in sorting out the etiology of hemodynamic instability. This author uses what he calls the “four-quadrant approach” to hemodynamic management for complex surgeries (Fig 1). This type of approach has a long history in cardiothoracic anesthesia, having first been described by Arkin, Benumof, and Saidman in 1977, for decision-making in separation from cardiopulmonary bypass. Because it invokes some important basic physiologic principles, it is useful for teaching as well. The patient's mean arterial pressure (MAP) is plotted on the y axis, and the stroke volume index (SVI) on the x axis. The approach can be used for GDFT focused on particular hemodynamic goals, in which stroke volume index (SVI) and blood pressure goals are set with a target zone in the center of the plot. In addition, it can facilitate problem solving; when a patient's hemodynamics stray out of the target zone, each quadrant indicates a certain family of hemodynamic aberrations that prompt corresponding management. For example, if a patient has high blood pressure and high SVI during surgery (zone 1), then hyperdynamic states, such as light anesthesia and catecholamine excess, would come to mind. This would prompt investigation and appropriate management. Importantly, volume responsiveness as assessed by dynamic parameters (stroke-volume variation, pulse-pressure variation) or PLR is used primarily in zone 3, in which both SVI and blood pressure are low. Fluid responsiveness assessment is used as a tool, not an endpoint, to determine if the shock state is cardiogenic (fluid responsiveness unlikely) or from hypovolemia (fluid responsiveness likely). This author believes that this approach is consistent with how many cardiothoracic anesthesiologists and intensivists approach hemodynamic problems, and that it facilitates GDFT by prompting defined hemodynamic goals and therapies.

Zaky et al., with their well-designed, well-executed study, have demonstrated clearly that volume responsiveness, as a single, isolated metric, is unlikely to be useful to either predict or prevent AKI in perioperative cardiac surgical patients. Used in clinical context, however, along with indices such as stroke volume and oxygen delivery, fluid responsiveness metrics, such as PLR and dynamic parameters, may be very useful in GDFT and clinical problem solving.