Hemodynamic stability and maintenance of cerebral perfusion pressure are crucial to the treatment of patients with intracranial pathology (1). In patients undergoing brain surgery, preoperative fasting, induction of general anesthesia, diuretics, and intraoperative bleeding may decrease intravascular volume, blood pressure, and cerebral perfusion pressure, as well as compromise perfusion of other organs.
Conventional hemodynamic variables, such as blood pressure, heart rate (HR), central venous pressure (CVP), and even pulmonary artery occlusion pressure (PAOP), are insensitive and sometimes misleading in the assessment of circulating blood volume (2). Measuring left ventricular (LV) end-diastolic area by transesophageal echocardiography (TEE), although considered to be the clinical “gold standard” for the estimation of preload, is limited to a small number of patients and is not routinely used in most operating rooms (3).
Previous experimental and clinical studies have shown that during controlled mechanical ventilation preload can be successfully assessed by analysis of changes in systolic blood pressure (SBP) during inspiration (4,5). The main hemodynamic effect of the increased intrathoracic pressure during positive pressure ventilation is a transient decrease in venous return and right ventricular filling, leading eventually to a decrease in LV stroke output and a decrease in the SBP. This transient decrease in SBP has been termed Δ down (dDown), and is measured as the difference between the SBP during a short apnea period and the minimal value of the SBP during the mechanical respiratory cycle. The dDown and the systolic pressure variation (SPV), which is the difference between the maximal and minimal SBP values during one mechanical breath, were found to correlate with the degree of blood loss and the decrease in cardiac output (CO) during experimental and clinical graded hemorrhage (6,7). They were also found to correlate with echocardiographic estimates of LV filling and with the response of the CO to volume loading (8,9). Indeed, the dDown and the SPV have a unique value in that they reflect volume responsiveness, namely the slope of the LV function curve rather than being static variables as are filling pressures (4).
The recently introduced PiCCO monitor (PULSION, Munich, Germany) offers a similar method for measuring the effects of mechanical ventilation on cardiac performance. This system measures the CO by transpulmonary thermodilution, the injection of cold saline through a central venous catheter, and mea-surement of the thermodilution curve by a thermistor-tipped catheter placed in the femoral or axillary artery. Further analysis of the thermodilution curve gives information on the cardiac filling volumes, namely the global end diastolic volume (GEDV) and the intrathoracic blood volume (ITBV), and on extravascular lung water (EVLW). After the initial measurement, a refined algorithm of the pulse contour method continuously measures CO and stroke volume (SV) (10,11). In addition, the monitor continuously displays the SV variation (SVV), which is the percentage of change between the maximal and minimal SVs divided by the average of the minimum and maximum over a floating period of 30 s. Thus, the SVV, like the SPV, reflects the ventilation induced changes in the LV stroke output and may serve as a variable that continuously assesses the adequacy of preload.
The aim of this study was to assess whether SVV can serve as a predictor of fluid responsiveness in patients undergoing elective brain surgery and to compare its predictive value to the CVP and to commonly measured hemodynamic variables.
After we obtained ethics committee approval and informed consent, 15 patients undergoing surgery for brain tumor removal or clipping of an intracerebral aneurysm were enrolled in the study. Patients with known significant cardiac or respiratory disease and patients who were not in cardiac sinus rhythm were excluded.
All patients were given preoperative antibiotics, dexamethasone, and anticonvulsants. Premedication consisted of oral diazepam 0.1 mg/kg 1 h before arrival in the operating room. Anesthesia was induced with IV fentanyl 2–3 μg/kg and thiopental 2–4 mg/kg. Vecuronium bromide 0.1 mg/kg IV was used to facilitate endotracheal intubation, and ventilation was controlled to an end tidal CO2 of 28–32 mm Hg, keeping a constant tidal volume of 10 mL/kg. Fentanyl 1–2 μg/kg, thiopental 2 mg/kg, and labetalol 5–10 mg IV were added before head frame positioning, and anesthesia was maintained by using additional fentanyl and 0.5%–0.8% isoflurane in a mixture of oxygen and air. Labetalol in boluses of 5–10 mg was used to control excessive increases in blood pressure, and mannitol 0.5 g/kg IV was administrated during craniotomy.
After anesthesia induction and before the beginning of surgery, an 8F central venous catheter was inserted in the right internal jugular or subclavian veins, and a 4F arterial thermodilution catheter (PULSION, Munchen, Germany) was inserted in the femoral artery. CO was measured in triplicates by injecting 0.2 mL/kg of 2–5°C normal saline through the CVP catheter, the thermodilution curve being measured by the arterial catheter that was connected to the PiCCO monitor. SBP, HR, SV, and SVV were continuously measured. CVP was continuously measured by using a transducer calibrated to the level of the right atrium by using an AS3 monitor (Datex/Ohmeda, Helsinki, Finland).
All patients were given 500 mL of normal saline IV during the induction of anesthesia and were then maintained with 1 mL · kg−1 · h−1 of normal saline solution. Additional fluids were given when deemed necessary by the attending anesthesiologist. After a period of 5 min of stable HR, BP, CVP, continuous CO, SV, and SVV measurements, volume loading was performed in increments of 100 mL of colloid solution (6% hydroxyethylstarch) for 2 min for each volume loading step (VLS). The hemodynamic variables were recorded again 1 min after the end of the infusion, and responsive VLS (increase in SV by at least 5%) and nonresponsive VLS (no change, or increase in SV by <5%) were identified. In each patient, successive VLSs were performed until a nonresponsive VLS was reached. If a nonresponsive VLS was reached after the first 100 mL of fluid infusion, one more VLS was performed and the hemodynamic variables recorded. No changes in anesthetic management were made during the study.
No VLSs were performed if stable baseline hemodynamic variables were not obtained for 5 min as a result of bleeding or necessary changes in anesthetic drugs dosage.
All hemodynamic variables were analyzed as continuous variables and expressed as the mean ± sd. To determine whether hemodynamic variables changed in relation to volume loading, differences between values before and after each VLS were compared between responsive and nonresponsive VLSs by using a two-tailed t-test. The correlation between changes in SV and changes in hemodynamic variables was assessed by using Pearson’s correlation. To assess the ability of different hemodynamic variables to discriminate between positive (>5% increase in SV) and negative (<5% increase in SV) response to fluid challenge, receiver operating characteristic (ROC) curves were generated for HR, SBP, CVP, and SVV, varying the discriminating threshold of each variable. The area under the ROC curve for each variable was calculated and compared (9,12). Values for each area can be between 0 and 1. A value of 0.5 indicates that the screening measure is no better than chance, whereas a value of 1 implies perfect performance. In our study, the area under the ROC curve represented the probability that a random pair of responsive and nonresponsive VLSs would be correctly ranked by the hemodynamic variable measurement.
Fifteen patients, 6 men and 9 women, ages 55 ± 15 yr, participated in the study. The indication for surgery included removal of a brain tumor in 10 patients, and clipping of an intracerebral aneurysm in 5. A total of 140 VLSs were performed. In 70 VLSs, an increase in SV of more than 5% was induced by the fluid challenge (responsive), while in 70 the SV increased by <5% after the procedure (nonresponsive). The mean number of VLSs performed before reaching unresponsiveness of the SV was 2 ± 1. In 25 of 140 VLSs, unresponsiveness was reached after the first fluid loading, a second loading was performed and in only one case did it increase SV further. The data from the additional loading were not included in the statistical analysis.
Responsive and nonresponsive VLSs differed significantly in their pre-VLS values of SBP, SV, and SVV, but not in the values of HR and CVP (Table 1). Similar relations were also found in the changes of these variables after fluid loading (Table 1). Statistically significant correlations were found between the change in SV after VLS and the values of SBP, SV, and SVV before fluid loading. Also significant was the correlation between the change in SV and the changes in these variables after VLS (Table 2). No correlation was found between the changes in SV and the values of the CVP and HR before loading or the change in these variables after VL (Table 2).
The overall performance of preload variables in predicting the responsiveness of the SV to VLS was evaluated by constructing ROC curves (Fig. 1). The areas under the ROC curves were 0.493 for the CVP (95% confidence interval [CI]: 0.397 to 0.590), 0.593 for the HR (95% CI: 0.443 to 0.635), 0.729 for the SBP (95% CI: 0.645 to 0.813), and 0.870 for the SVV (95% CI: 0.809 to 0.903). The area for SVV was statistically more than those for CVP (P = 7 × 10−10), HR (P = 5.7 × 10−10), and SBP (P = 4.3 × 10−3). The optimal threshold values given by ROC analysis were 9.5% for SVV. Thus, if a patient had a SVV value of more than 9.5%, he was very likely to be responsive to a subsequent volume load by increasing his SV by more than 5% with a sensitivity of 78.6% and a specificity of 93%.
To maximize cerebral perfusion, it is imperative that optimal preload conditions be achieved in patients undergoing neurosurgery (13). The effect of anesthetics, as well as the frequent use of β-adrenergic blocking drugs, limits the value of BP and HR for volume status assessment. Urine output is also of limited value because of the frequent use of mannitol and loop diuretics. Measurement of cardiac filling pressures, namely CVP and PAOP, may be misleading because of changes in heart compliance, high airway pressures, valvular regurgitation, technical failure, etc. (14,15). A more accurate method for preload assessment is based on the measurement of LV end-diastolic area by using TEE. This method, however, is limited to a small number of patients, and cannot be used continuously for long periods of time.
Observing the interactions between the respiratory and the cardiac systems during mechanical positive pressure ventilation prompted the use of SPV as a predictor of fluid responsiveness. The SPV and its dDown component (the difference between the minimal SBP value during a mechanical breath and the SBP value during apnea) were shown to be sensitive indicators of hypovolemia in ventilated dogs (5,6) and in humans subjected to graded hemorrhage (7,16). Furthermore, both Coriat et al. (8) and Tavernier et al. (9) demonstrated that the SPV was a more accurate indicator than the PAOP for fluid responsiveness in patients undergoing vascular surgery or in patients with septic shock, respectively. Nevertheless, although found clinically useful (17), and although a recent review recommended its use in neurosurgical patients (13), the technique has not yet been made commercially available. Although the SPV may be approximated from the arterial waveform, accurate measurement is difficult, unless a cursor is used or the arterial trace is printed out and measurements made off-line. The measurement of the dDown is even more complex as it necessitates the introduction of an apnea period, making the technique cumbersome for routine clinical practice.
The recently introduced PiCCO monitor offers the option of continuous measurement of the effects of mechanical ventilation on the cardiac function. In this system, the thermodilution curve is measured by a thermistor-tipped catheter placed in the femoral or axillary artery after an injection through a catheter in the jugular or subclavian veins. Further analysis of the thermodilution curve, and calculation of the mean transient time of the indicator and its down sloping, allows for the calculation of other variables. These include the GEDV of the heart chambers, the ITBV combination of the GEDV and the pulmonary blood volume, and EVLW. After CO measurement by using transpulmonary arterial thermodilution technique, SV is measured continuously by using pulse contour analysis. This method is based on the existing relationship between the area under the BP curve and the associated SV (18–20). The PiCCO device, by using advanced algorithms that take into account on-line aortic compliance, and using a transpulmonary (arterially) measured CO, allows continuous measurement of SV, including continuous display of its variability, the SVV, which is the percent of change of the SV. As already indicated, this variability during mechanical ventilation may reflect volume responsiveness, because the decrease in the venous return during the mechanical breath will greatly affect SV during hypovolemia (large SVV) but not during hypervolemia or congestive heart failure (small SVV).
The results of the present study confirm the ability of the SVV to predict fluid responsiveness to even a small volume loading of 100 mL of 6% hydroxyethylstarch given for two minutes. A SVV value that is equal to or more than 9.5% will predict positive fluid responsiveness, defined as an increase in SV by 5% or more, with a sensitivity of 78.6% and a specificity of 93%. By using ROC analysis, it was demonstrated that not only was the predictive value of the SVV superior to the predictive value of CVP and HR, but also that the area under the curves indicates these screening measures are not better than random chance. The ROC analysis also demonstrated the superiority of SVV over SBP as a predictor of fluid responsiveness (P = 4.3 × 10−3).
The ability of the SVV variable to predict the responsiveness to such a small volume load and the continuous measurement of both SVV and SV are of utmost clinical importance. Because the relationship between the cardiac LV output and preload in a defined contractility are not linear, the ability to predict whether the heart will augment its function after fluid administration, or whether inotropes are needed is crucial. This ability may minimize unnecessary volume loading, which is significant mainly in patients with cardiac or renal dysfunction. Moreover, the continuous measurement makes the assessment of LV function curve easier. Instead of the time consuming maneuver of CO and PAOP measurement after each step of fluid loading using the PAC, the SV can be continuously plotted against the SVV and a dynamic on-line estimation of cardiac function can be achieved.
The relation between cardiac function and mechanical ventilation was recently studied by Denault et al. (21). Using TEE derived variables including LV end diastolic area, and stroke areas, these authors claim that changes in the arterial blood pressure during mechanical ventilation occurred in a different time phase and magnitude than changes in echocardiographically-measured SV. The authors concluded that changes in BP during mechanical ventilation reflect changes in intrathoracic pressure and not changes in LV hemodynamics so that the SPV cannot be used as an indicator for preload. However, their study was conducted in a small number of patients with both closed and open chests, and the determination of beat-to-beat cardiac SV by TEE may have been of limited accuracy, as a result of factors such as the vertical movement of the heart during inspiration. More recent work has been done in patients with acute respiratory failure, in whom application of positive end expiratory pressure caused significant changes in CO that correlated with the baseline pulse pressure change during mechanical ventilation, a variable that is very close to the SPV (22). In contrast to the findings of Denault et al. (21), other authors have found that mechanical ventilation induced changes in SV, with an early augmentation followed by later decrease (23).
Although the ability of SVV to accurately predict fluid responsiveness was demonstrated, a major disadvantage of the study is that other preload variables such as SPV, PAOP, or LV end diastolic area, were not measured simultaneously with the SVV. The PiCCO-derived preload variables, i.e., ITBV and the GEDV, were not measured either. The injection of indicator through a CVP line is difficult during brain surgery when the neck is in the operative field. Such a comparison might have shown that the SVV overestimates hypovolemia, because of the inclusion of the Δ up as a component of the SVV. The Δ up, which is the difference between the maximal SBP during the mechanical breath and the SBP during apnea, reflects augmentation of the SV caused by increased LV preload during inspiration. This variable might be prominent mainly in patients with LV failure, because of the afterload-reducing effect of inspiration induced by the pressure gradient between the chest and the abdomen during positive pressure ventilation (5). Other limitations of the present study are that the study protocol was performed by the anesthesiologist treating the patient without any blinding, the fact that multiple measurements were performed in the same patients, and the need for an arbitrary definition of responsiveness. Even with all these methodological limitations, and although the PiCCO monitoring system is more invasive and expensive than simple SPV assessment, it gives the benefits of continuous numerical SVV as well as continuous SV measurements and other variables not reported in this study. In addition, thermodilution derived variables (CO, ITBV, GEDV, and EVLW) may be beneficial in the treatment of high-risk patients.
We would like to thank Dr. B. Tavernier from the Department d’Anesthesie-Reanimation 2, Hopital Claude Huriez, CHU de Lille, France, for his help in the statistical analysis.
1. Guidelines for cerebral perfusion pressure. J Neurotrauma 1996; 13: 693–7.
2. Messinger G, Segal E, Perel A. Monitoring of hypovolemia. Curr Opin Anaesthesiol 1993; 6: 393–9.
3. Schiller NB, Shah PM, Crawford M, et al. Recommendation for quantitation of the left ventricle by two-dimensional echocardiography: American Society of Echocardiography Committee on Standards, Subcommittee on Quantitation of Two-Dimensional Echocardiography. J Am Soc Echo 1989; 2: 358–67.
4. Perel A. Assessing fluid responsiveness by the systolic pressure variation in mechanically ventilated patients. Anesthesiology 1998; 89: 1309–10.
5. Pizov R, Ya’ari Y, Perel A. The arterial pressure waveform during acute ventricular failure and synchronized external chest compression. Anesth Analg 1989; 68: 150–6.
6. Perel A, Pizov R, Cotev S. The Systolic pressure variation is a sensitive indicator of hypovolemia in ventilated dogs subjected to graded hemorrhage. Anesthesiology 1987; 67: 498–502.
7. Rooke GA, Schwid HA, Shapira Y. The effect of graded hemorrhage and intravascular volume replacement on systolic pressure variation in humans during mechanical and spontaneous ventilation. Anesth Analg 1995; 80: 925–32.
8. Coriat P, Vrillon M, Perel A, et al. A comparison of systolic blood pressure variations and echocardiographic estimates of end-diastolic left ventricular size in patients after aortic surgery. Anesth Analg 1994; 78: 46–53.
9. Tavernier B, Makhotine O, Lebuffe G, et al. Systolic pressure variation as a guide to fluid therapy in patients with sepsis-induced hypotension. Anesthesiology 1998; 89: 1313–21.
10. Godje O, Hoeke K, Lichtwarck-Aschoff, et al. Continuous cardiac output by femoral arterial thermodilution calibrated pulse contour analysis: comparison with pulmonary arterial thermodilution. Crit Care Med 1999; 27: 2407–12.
11. Godje O, Hoeke K, Lamm P, et al. Continuous, less invasive, hemodynamic monitoring in intensive care after cardiac surgery. Thorac Cardiovasc Surg 1998; 46: 242–9.
12. Hanley JA, McNeil BJ. A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology 1983; 148: 839–43.
13. Stone DJ. Physiology and monitoring of intravascular volume status in the neurosurgical patient. Anesth Analg 1999; 88 (S-IARS):98–103.
14. Mark JB. Central venous pressure monitoring: clinical insights beyond the numbers. J Cardiothorac Vasc Anesth 1991; 5: 163–73.
15. Connors AF, Speroff T, Dawson NV, et al. The effectiveness of right heart catheterisation in the initial care of critically ill patients. JAMA 1996; 276: 889–97.
16. Ornstein E, Eidelman LA, Drenger B, et al. Systolic pressure variation predicts the response to acute blood loss. J Clin Anesth 1998; 10: 137–40.
17. Stoneham MD. Less is more: using systolic pressure variation to assess hypovolemia. BJA 1999; 83: 550–1.
18. Sakka SG, Reinhart K, Meier-Helmann A. Comparison of pulmonary artery and arterial thermodilution cardiac output in critically ill patients. Intensive Care Med 1999; 25: 843–6.
19. Buhre W, Weyland A, Kazmaier S, et al. Comparison of cardiac output assessed by pulse-contour analysis and thermodilution in patients undergoing minimally invasive direct coronary artery bypass grafting. J Cardiothorac Vasc Anesth 1999; 13: 437–40.
20. Godje O, Thiel C, Lamm P, et al. Less invasive continuous hemodynamic monitoring during minimally invasive coronary surgery. Ann Thorac Surg 1999; 68: 1532–6.
21. Denault AY, Gasior TA, Gorscan J 3rd>, et al. Determinants of aortic pressure variation during positive pressure ventilation in man. Chest 1999; 116: 176–86.
22. Michard F, Chemla D, Richard C, et al. Clinical use of respiratory changes in arterial pulse pressure to monitor the hemodynamic effects of PEEP. Am J Respir Crit Care Med 1999; 159: 935–9.
© 2001 International Anesthesia Research Society
23. Jardin F, Delorme G, Hardy A, et al. Reevaluation of hemodynamic consequences of positive pressure ventilation: emphasis on cyclic right ventricular afterloading by mechanical lung inflation. Anesthesiology 1990; 72: 966–70.