Moving Points in Nephrology Advances in Critical Care for the Nephrologist: Hemodynamic Monitoring and Volume Management

The monitoring of physiologic variables is an integral part of the diagnosis and management of the critically ill patient. Restoration of tissue perfusion and oxygen delivery is the ultimate goal for any state of circulatory collapse. Insight into a patient's intravascular volume status and cardiac performance, particularly in the early stages of shock, can help guide management and potentially change outcome. In the past 30 years, various bedside monitoring techniques and indices have been developed in an effort to determine and optimize a patient's cardiac performance. This article reviews the physiologic parameters that best predict intravascular volume status and volume responsiveness. We examine the controversies surrounding the pulmonary arterial catheter and describe the less invasive methods of measuring cardiac performance. Assessing Volume Status in the Intensive Care Unit ume expansion, in an effort to improve cardiac output (CO) and augment tissue perfusion, is a common therapeutic goal in the hemodynamically unstable patient. Volume expansion is not without adverse effects, including pulmonary edema, worsening gas exchange, and hyperchloremic acidosis (1). It has been shown that not all patients in circulatory collapse respond to volume with improved cardiac function (1– 6). Moreover, patients with capillary leak may have total body volume overload and yet still benefit from augmented intravascular volume. Discriminating between patients who will benefit from volume expansion and those for whom an inotrope or vasopressive agent will better augment perfusion can limit the adverse effects of volume overload. Central venous pressure (CVP) and pulmonary artery occlusion pressure (PaOP; or "wedge pressure") have traditionally been used to estimate preload and intravascular volume status. These pressure-derived preload values, or filling pressures, have been central to the management of fluid resuscitation and titration; however, numerous studies have challenged the notion that these indicators accurately predict volume status. In fact, more than a dozen studies that examined various patient populations (sepsis, perioperative cardiovascular surgery, trauma, and other critical illnesses) all failed to demonstrate a correlation among CVP/PaOP, volume status, and cardiac performance (1,3–12). The Frank-Starling curve shown in Figures 1 and 2 illustrates how filling pressures can lead to inaccurate and often misleading interpretation of intravascular volume status. Figure 1. Patient A has a steeper Starling curve than patient B. Although patients A and B have the same initial preload value and although the patients have identical changes in preload (i.e., fluid bolus), patient A has a greater increase in stroke volume than patient B. Patient A is said to be "volume responsive." SV, stroke volume. Figure 2. Patient C is on the steep portion of the curve. Patient D is on a flat portion. Identical changes in preload (i.e., fluid bolus) result in different stroke volumes. Pressure-derived preload values do not identify a position of or place on the Starling curve and, therefore, poorly predict whether volume will improve hemodynamics. For example, a low CVP might prompt the clinician to give a fluid bolus; however, because of the patient's position on the curve, volume will not result in an increase in the CO and may even result in pulmonary edema. Conversely, a patient can have "high" filling pressures yet still be on the vertical portion of the curve. Traditionally, an elevated CVP or PaOP would trigger an order to diurese, when, in actuality, the patient's physiology would benefit from volume enhancement (1,4,8). CVP and PaOP are also poor indicators of cardiac preload because they fail to reflect cardiac volume in the setting of reduced ventricular compliance (4,13). A noncompliant, "stiff" heart may generate high filling pressures even in the setting of underfilled ventricles. The CVP or PaOP may be elevated, yet the patient's CO would improve with volume infusion (8). Finally, external forces, including positive end expiratory pressure, abdominal pressures, and vascular compliance, alter the relationship between filling pressures and end diastolic volume (14). In a study by Osman et al. (8), CVP, PaOP, and cardiac index (CI) were measured before and after a challenge of 6% hydroxyethyl starch infusion. Volume responsiveness was defined as an increase in CI by 15%. Preinfusion values of CVP were not statistically different between responders and nonresponders (Figure 3). The preinfusion values of PaOP, although statistically different between the two groups, had a great deal of overlap; therefore, no threshold value to predict volume responsiveness could be defined (Figure 4) (8). Even when combining low CVP and low PaOP values, the filling pressures still failed to predict whether volume would enhance CO (8). Figure 3. Preinfusion CVP ( , individual values; •, mean values) of responders (R) and nonresponders (NR). Reprinted from Osman et al. (8), with permission. Figure 4. Preinfusion PaOP ( , individual values; •, mean values) of responders (R) and nonresponders (NR). Reprinted from Osman et al. (8), with permission. When examining the patterns shown in Figures 3 and 4, there is certainly a trend suggesting that patients with low filling pressures correspond to a response of fluid challenge. In fact, if one were to take a large population of patients and gather an aggregate of filling pressures, the low CVP and PaOP would likely predict volume responsiveness among this large group; however, when caring for the individual patient, these filling pressures do not predict with enough accuracy or reliability that a low value translates to volume responsiveness. Not only are the absolute values of CVP and PaOP poor surrogates for volume responsiveness, but also several studies have shown that changes in CVP and PaOP after a fluid challenge do not correlate with changes in CO (9,13,15,16). Kumar et al. (13) demonstrated this concept definitively in a study of healthy volunteers. In this study, cardiac contractility failed to respond in any consistent or predictable manner when compared with the changes in CVP or PaOP after a 3-L saline infusion. The authors concluded that filling pressures, both absolute values and trends after volume infusion, failed to predict volume responsiveness in this population. In the landmark study of early goal-directed therapy by Rivers et al. (17), CVP of 8 to 12 mmHg was one of the resuscitation end points. The American College of Critical Care Medicine guidelines for hemodynamic support of patients with sepsis also use CVP and PaOP to define resuscitation goals (6), yet the predefined increase in CI in the study by Osman et al. (8) was seen after fluid bolus in only 46% of patients whose CVP was <12 mmHg and 54% of patients whose PaOP was <12 mmHg. These findings are congruous with the results of many other studies and strongly question the accuracy and utility of CVP and PaOP as predictors of preload and targets of volume therapy (3,5,13,18,19). Although CVP and PaOP may predict intravascular volume overload in certain clinical situations and PaOP is used in the definition of acute respiratory distress syndrome (ARDS), these values are just as likely to be misleading. A clinical example is the patient with bacterial pneumonia who is hypotensive and hyperlactemic and requires high levels of positive end expiratory pressure to promote gas exchange. Although this patient's PaOP may be 19 mmHg (a level defined as consistent with cardiogenic pulmonary edema), diuresis would only worsen tissue perfusion (20). The clinician should use caution when making interventions on the basis of a single parameter whose reliability is of great uncertainty. Additional parameters that have proved to be moderately useful in quantifying intravascular volume status include left ventricular end diastolic volume and area determined by echocardiogram. Although these cardiac dimensions are more direct assessments of intravascular volume, they still have limited ability to predict whether changes in preload will affect hemodynamics (1). Cardiomyopathy, valvular disease, and echocardiogram variability adversely affect the reliability of these measurements. In contrast to PaOP, CVP, and cardiac dimensions, the so called "static" markers, "dynamic" markers are those that use variations in either stroke volume or arterial pressure because the physiologic effects of respiratory variation more reliably predict volume responsiveness. In patients who are on positive pressure mechanical ventilation, inspiration causes a reduction in right ventricular preload as a result of compression of the vena cava, whereas right ventricular afterload is increased because of increased alveolar pressures. The result is a reduction in right ventricular ejection during inspiration. Because of the transit time of blood flow from the right to the left side of the heart (approximately 2 s), a fall in stroke volume and BP is seen during expiration (Figure 5). Because the hemodynamic effects that are induced by mechanical ventilation are exaggerated in the hypovolemic patient, the greater the variation, the more likely the patient's hemodynamics will improve with volume. These dynamic markers can be seen at the bedside as variations in arterial pressure, pulse pressure, or stroke volume (21,22). A graphic depiction and definitions of these dynamic markers are shown in Figure 5 and Table 1 (21). Figure 5. Dynamic markers. (A) Systolic pressure (SP) variation and decrease in arterial pressure during expiration ( Down). (B) Pulse pressure (PP) variation. Further definitions provided in Table 1. Table 1. The dynamic markers Dynamic Marker Definition SPV The difference between the maximum systolic pressure and minimal systolic pressure after a breath Down The decrease in arterial pressure during expiration PPV The maximum minus the minimum of the pulse pressure divided by the average of the two over a mechanical breath SVV The percentage change between the maximum stroke volume and minimum stroke volume over a designated interval a PPV, pulse pressure variation. Perel et al. (23) performed one of the first studies that examined systolic pressure variation as a predictor of volume responsiveness in a dog model of hemorrhagic shock. Variation in systolic pressure correlated with degree of hemorrhage better than CVP, heart rate, or mean arterial pressure. Tavernier et al. (5) examined dynamic and static variables including systolic pressure variation, decrease in arterial pressure during expiration, PAOP, and left ventricular end diastolic area index in patients with sepsis. Hemodynamic values were recorded before and after a volume challenge. Responders were predefined as having an increase in CI by 15%. The decrease in arterial pressure during expiration was the best predictor of improved cardiac function with volume (receiver operating characteristic 0.97) compared with PaOP (receiver operating characteristic 0.67). Marx et al. (19) compared stroke volume variation (SVV) with PaOP and CVP as a predictor of volume responsiveness in patients with sepsis. SVV predicted the cardiovascular response to fluid, whereas the filling pressures were no better than chance in anticipating hemodynamic change (10). With the use of a similar study design, dynamic markers have been tested among various patient populations, including neurosurgical patients (22), coronary artery bypass graft patients (18,24), and patients with acute lung injury (25). To our knowledge, all but one study have shown that the dynamic markers predict with reasonable accuracy whether a volume infusion will result in improved cardiac function (5,10,18,24,25). There are some limitations to the use of dynamic markers. For example, dynamic parameters are affected by varying tidal volumes and therefore require a well-sedated, mechanically ventilated patient to ensure accuracy. Arrhythmias, particularly atrial fibrillation, will also affect the precision of SVV analyses. Finally, alterations in myocardial contractility as a result of titration of inotropic or vasopressive agents can affect the accuracy of these dynamic markers (4,18,19). It should also be emphasized that the dynamic markers do not provide information of absolute intravascular volume status (i.e., a low SVV does not suggest volume overload) but, rather, predict whether a patient will benefit from volume infusion. Although dynamic measures are more accurate at predicting volume responsiveness, as a practical matter, a clinician will often try a volume challenge in a patient who is hypotensive or oliguric. If volume produces no effect on CO, then an inotrope may be started. Because BP is often being modified with vasopressors and oliguria may be slow to resolve, this underscores the need for bedside monitoring of CO, as we will discuss in the next section. Hemodynamic Devices: The Pulmonary Artery Catheter Developed in the 1970s, the pulmonary artery catheter (PAC) provided bedside hemodynamic parameters unlike other devices before its time. The PAC provides measurements including CVP, right atrial and right ventricular pressures, pulmonary artery pressures, PaOP, mixed venous saturation (SvO2), and CO. Systemic and pulmonary vascular resistances are calculated from these values (26). The value of the PAC has been a subject of debate in the past two decades. Proponents of its use argue that the PAC provides data necessary for the appropriate diagnoses and treatment of cardiovascular failure. The PAC is an accurate technique for obtaining CO, and it continues to be the gold standard for CO measurement at the bedside. In addition, pulmonary pressures obtained from the PAC can be used to evaluate and modify pharmacologic treatment for pulmonary hypertension. Last, the SvO2 reflects the overall balance among oxygen supply, demand, and use by the body's tissues and is used, along with lactate levels, to guide resuscitation. Despite the real-time, hemodynamic parameters provided by the PAC, there are still no data supporting its use in the care of the critically ill patient. Moreover, there are other means of obtaining similar data yet in a less invasive manner. For example, unlike SvO2, the central venous saturation (ScvO2) is drawn from the superior vena cava via a central catheter and does not require a PAC. The ScvO2 has gained increasing acceptance since its use in the early goal-directed therapy algorithm by Rivers et al. (17) and can be measured in a continuous manner. Although ScvO2 values are higher than SvO2, the gradient is consistent and therefore ScvO2 is a reliable surrogate for SvO2 (27). (On average, ScvO2 is approximately 5% higher than SvO2. This difference, however, can vary among individual patients [27].) Finally, less invasive devices that provide hemodynamic parameters similar to that of the PAC are now available. These less invasive devices as well as data on the usefulness of the PAC are described in the sections to follow. As with any invasive device, complication rates are an important consideration in a benefit-to-risk analysis. The process of obtaining initial venous access with a central venous catheter and PAC are similar; therefore, complication rates at the stage of catheter insertion are comparable. (The risk for mechanical complications from catheter insertion can be diminished by the use of ultrasound guidance. When compared with the standard landmark technique, live ultrasound guidance has been shown to decrease arterial puncture and other complications [28].) The PAC carries additional risks given its advancement and residence in the right heart and pulmonary vasculature. Significant dysrhythmias, heart block, pulmonary artery rupture, knotting, and increased risks for infection over a simple central venous catheter are examples of these potential adverse events (29,30). The hemodynamic parameters and adverse effects of the PAC and central venous catheter are compared in Table 2 (30). Table 2. A comparison of the pulmonary artery catheter and the central venous catheter Parameter PAC Central Venous Catheter Relative Accuracy of Each Parameter Obtainable measurements