Fluid therapy to maintain optimal cardiac output is an essential part of the management of anaesthesia for cardiac surgery.1 Because fluid loading is generally considered as a first-line therapy for treating patients with low cardiac output, it is important to predict whether it leads to increases in cardiac output.2 Numerous studies have shown that dynamic preload indices such as pulse pressure variation (PPV) and stroke volume variation (SVV) are superior in predicting fluid responsiveness compared with the traditional use of cardiac filling pressures in mechanically ventilated patients,3–11 and therefore, such dynamic indices have been widely used in various clinical settings.
PPV and SVV calculations are based on cyclic changes in intrathoracic pressure produced by mechanical ventilation.12 Therefore, any alteration in pressure transduction from the lungs to the thoracic cardiovascular system, which is also related to opening of the thorax,13,14 could affect the accuracy of PPV and SVV as predictors of fluid responsiveness. Several previous studies have investigated the usefulness of dynamic preload indices for assessment of fluid responsiveness under open-chest conditions; however, the results remain controversial.1,12,13,15–17
Interestingly, two studies conducted in spontaneously breathing patients reported that PPV calculated during forced inspiration or the Valsalva manoeuvre18 can overcome the limitations of PPV associated with low tidal volumes. Considering that the effect of low tidal volume on dynamic preload indices might be similar to that of open-chest conditions, mainly characterised as reduced effects of intrathoracic pressures on the thoracic cardiovascular structures,1,19,20 applying high tidal volumes in open-chest conditions would restore the predictability of PPV for fluid responsiveness through increased effects on alveolar pressure. Therefore, in this prospective observational study, we hypothesised that augmented PPV using a Valsalva manoeuvre can predict fluid responsiveness in open-chest conditions.
Materials and methods
This prospective, observational study (IRB No. 2014-07-005) was approved by the Institutional Review Board of Samsung Medical Centre, Seoul, Korea on 11 September 2014, and registered at ClinicalTrials.gov (NCT02457572). Written informed consent was sought from each patient who was asked to participate.
From October 2014 to June 2015, we enrolled adult patients aged 20 to 80 years scheduled for elective off-pump coronary arterial bypass grafting (OPCAB). Exclusion criteria were preoperative left ventricular ejection fraction less than 35%, right ventricular dysfunction; valvular heart disease, preoperative cardiac arrhythmia, intracardiac shunts, severe renal or liver disease, chronic obstructive pulmonary disease, moderate to severe pulmonary hypertension with mean pulmonary arterial pressure more than 35 mmHg and presence of bullous lung disease.
When each patient arrived in the operating room, routine monitoring was applied including 5-lead electrocardiography with ST-segment analysis, pulse oximetry and noninvasive blood pressure, and intravenous midazolam 1 mg was administered for premedication. A radial artery catheter was inserted under local anaesthesia before induction of anaesthesia. Anaesthesia was induced with intravenous etomidate 0.2 mg kg−1 and sufentanil 0.5 to 2 μg kg−1. Neuromuscular block was achieved with rocuronium 0.6 mg kg−1. After tracheal intubation, the lungs were mechanically ventilated with a tidal volume of 8 ml kg−1, an inspired oxygen fraction of 0.5 and no positive end-expiratory pressure in either closed or open-chest conditions. The respiratory rate was adjusted to maintain the partial pressure of end-tidal carbon dioxide at 4.7 to 5.3 kPa (35 to 40 mmHg). Anaesthesia was maintained with inhaled isoflurane targeting BIS values between 40 and 60. Muscle relaxation was maintained with a continuous infusion of vecuronium (8 to 10 mg h−1).
After induction of anaesthesia, a standard 7.5-French gauge pulmonary artery catheter (Swan-Ganz CCOmbo CCO/SvO2; Edwards Life Sciences, Irvine, California, USA) was introduced through an 8.5-French gauge two-lumen introducer (MAC; Arrow, Reading, Pennsylvania, USA) into the right internal jugular vein and connected to an analysis system (Vigilance II; Edwards Lifesciences) for cardiac output monitoring. A probe was also inserted for transoesophageal echocardiography. All pressure transducers were zeroed at the patient's mid-axillary level to ambient pressure. Following those procedures, mean arterial pressure (MAP), heart rate (HR), central venous pressure (CVP) and mean pulmonary artery pressure (MPAP) were displayed continuously.
After midline sternotomy, haemodynamic variables, including MAP, HR, CVP, MPAP, cardiac index (CI) and PPV, were measured or calculated before and 5 min after fluid loading. After measurement of baseline haemodynamic data, the Valsalva manoeuvre for PPV augmentation was performed. Three successive Valsalva manoeuvres were applied with a more than 1-min recovery interval between each manoeuvre, and mean PPV was calculated from these three values. Details of the Valsalva manoeuvre performance were as follows: ventilator mode was switched from volume-controlled to manual mode; the adjustable pressure-limiting valve of the ventilator was partially closed; and the rebreathing bag was manually held to maintain a nearly constant airway pressure of 30 cmH2O for the 10 s before volume loading (Fig. 1). When the Valsalva manoeuvre was completed, the ventilator mode was switched back to the volume-controlled mode. A balanced crystalloid solution (Plasma Solution-A Inj., CJ Pharma, Seoul, Korea) in a volume of 6 ml kg−1 of ideal body weight was infused through a central venous catheter for volume expansion within 10 min after the Valsalva manoeuvre.
Cardiac output was determined using the mean of three thermodilution measurements by a bolus injection of 10 ml of cold 0.9% saline. Each bolus injection was performed at the same point in the respiratory cycle (at end-expiration). The reproducibility of measuring cardiac output was regarded as acceptable if the lowest value was within 20% of the highest. CI was calculated as cardiac output/body surface area that was determined using the DuBois formula [body surface area = body weight (kg)0.425 × body height (m)0.725 × 0.20247].
Automatically calculated PPV (PPVauto) was recorded from the IntelliVue patient monitor (Intellivue MP 70: Philips Medical Systems, Böblingen, Germany). PPV was calculated automatically from the arterial pressure waveform using previously described algorithms21,22 (PPVauto) and displayed on the IntelliVue MP70 monitor in real time. We also calculated PPV manually (PPVmanual) at each study time point using the simultaneously recorded arterial pressure and airway pressure waveforms. Pulse pressure (PP) was defined as the difference between the systolic and diastolic arterial pressures. Maximal PP (PPmax) and minimal PP (PPmin) values were determined over a single respiratory cycle. PPVmanual was calculated as follows: PPVmanual = (PPmax - Pmin)/[(PPmax + PPmin)/2]. The measurements were repeated over three respiratory cycles and averaged. Augmented PPV during Valsalva manoeuvres (PPVVM) was also calculated using the same equation for PPVmanual over the inflation period of each individual Valsalva manoeuvre (Fig. 1).
All measurements were performed under open-chest, closed-pericardium conditions. The pleurae were closed. Before performing each Valsalva manoeuvre, we notified the surgeons and they discontinued patient manipulation for a brief period. Because our study was performed during the stable surgical period (just after midline sternotomy and immediately before or during internal thoracic artery exposure and separation from the chest wall), intravenous fluid infusion was restricted unless the patient was given intravenous drugs including anaesthetic agents. Regarding the use of intraoperative inotropic drugs or vasopressors, hypotension was treated using the standardised hospital protocol. Hypotension was defined as systolic arterial pressure less than 90 mmHg or MAP decrease more than 30% from baseline pressure. If HR was less than 75 bpm, ephedrine 5 mg was administered and if HR was greater than 75 bpm, phenylephrine 30 μg was given. However, vasoactive or inotropic drugs were not administered during the study period with the exception of a continuous infusion of intravenous nitroglycerine. In most patients, intravenous nitroglycerine (0.3 to 0.5 μg kg−1 min−1) was administered continuously at a stable dose throughout the study period. In addition, if severe haemodynamic instability requiring medical or surgical therapy developed during the study protocol, appropriate treatment was initiated and the patient was excluded from analysis.
Data are presented as number of patients (%), mean ± SD or median [interquartile range, IQR]. Percentage changes in CI according to fluid loading were used as the principal indicator of fluid responsiveness. Patients were classified as responders or nonresponders when increases in CI were at least 12% or less than 12% after volume loading, respectively. To test the ability of PPVauto, PPVmanual, PPVVM and CVP to predict fluid responsiveness, areas under the receiver operating characteristic (ROC) curves of the responders were calculated and compared using the Hanley–McNeil test. Briefly, the general interpretations of a test according to the value of the area under the ROC curves are as follows: area under the curve (AUC) = 0.5, no better than chance, a useless test with no prediction possible; AUC ranged from 0.6 to 0.69, a test with a poor predictability; AUC ranged from 0.7 to 0.79, a fair test; AUC ranged from 0.8 to 0.89, a test with a good predictability; AUC ranged from 0.9 to 0.99, an excellent test; AUC = 1.0, a perfect test with best possible prediction. A threshold value for each variable to discriminate fluid responsiveness was determined using the Youden index (Youden Index = sensitivity + specificity - 1) to maximise both sensitivity and specificity. To compensate for the dichotomous division of patients into fluid responders or nonresponders in ROC curve analysis, the relationship between change in CI and PPVs (PPVauto, PPVmanual and PPVVM) before fluid loading was assessed with Pearson's correlation.
Patient characteristics were compared between responders and nonresponders using the Student's t-test or Mann–Whitney U-test for continuous variables and the χ2 test or Fisher's exact test for categorical data. Haemodynamic variables were compared before and after fluid loading using the paired t-test or Wilcoxon signed-rank test.
To calculate sample size, MedCalc for Windows, version 15.6.1 (MedCalc Software, Ostend, Belgium) was used. We considered a ROC curve with an AUC at least 0.8 to indicate a clinically reliable predictor for fluid responsiveness. Therefore, considering that the null hypothesis was 0.5, at least 17 patients were required in each group to detect a 0.3 difference in the AUC for predicting fluid responsiveness with a significance level of 0.05 and a probability power of 90% when the number of responders was assumed to be similar to that of nonresponders. All statistical analyses were performed using SPSS 22.0 for Windows (IBM Corp., Armonk, New York, USA) and MedCalc for Windows, version 15.6.1. P value less than 0.05 was considered statistically significant.
Of the 89 patients scheduled for elective OPCAB who were screened, 52 patients were enrolled in the study and data from 49 patients were ultimately analysed (Fig. 2). After volume expansion, there were 21 fluid responders and 28 nonresponders. Patient characteristics were comparable between responders and nonresponders (Table 1). All included patients had normal preoperative right ventricular systolic function.
When applying the Valsalva manoeuvre, arterial pressure showed a slight initial rise and then gradually fell during the inflation period (Fig. 1). The lowest median [IQR] MAP during the inflation period of the Valsalva manoeuvre was 52 [49 to 53] mmHg (P < 0.001 vs. baseline), significantly lower than baseline MAP before the Valsalva manoeuvre [72 (67 to 75) mmHg]. However, arterial pressure recovered quickly with Valsalva manoeuvre release [68 (64 to 73) mmHg, P = 0.2 vs. baseline].
Changes in haemodynamic variables before and after volume expansion are summarised in Table 2. In fluid responders, CVP, MPAP and CI increased consistently after fluid loading. However, in nonresponders, fluid loading caused an increase in CVP, while a decrease in CI was observed (Table 2). Compared with nonresponders, responders had higher PPVauto [8 (6 to 11) vs. 5 (4 to 7)%, P = 0.003] and PPVmanual [8 (5 to 9) vs. 5 (4 to 6)%, P = 0.003], and lower CI [2.5 (0.5) vs. 3.0 (0.6) l min−1 m−2, P = 0.001] before fluid loading (Table 2, Fig. 3).
PPVVM before fluid loading significantly correlated with volume expansion-induced changes in CI [r = 0.6 (95% CI 0.4 to 0.7), P < 0.001, Fig. 4a]. However, baseline PPVauto or PPVmanual was not significantly correlated with the volume expansion-induced changes in CI [r = 0.2 (95% CI -0.1 to 0.5), P = 0.11 in PPVauto and r = 0.2 (95% CI -0.1 to 0.5), P = 0.12 in PPVmanual, Fig. 4b, c].
In the ROC curve analysis (Table 3, Fig. 5), PPVVM showed powerful predictability with an AUC of 0.88 (95% CI, 0.75 to 0.95, P < 0.001). A PPVVM value of 55% predicted fluid responsiveness with a sensitivity of 90.5% (95% CI 69.6 to 98.8) and a specificity of 78.6 (95% CI 59.0 to 91.7). Baseline PPVauto and PPVmanual also predicted fluid responsiveness with AUC of 0.76 and 0.75, respectively (Table 3, Fig. 5).
In the present study, we demonstrated that augmented PPV performed during a Valsalva manoeuvre reliably predicts fluid responsiveness under open-chest conditions. A threshold PPVVM value of 55% was able to discriminate fluid responders from nonresponders. Baseline PPV after sternotomy also predicted fluid responsiveness with a threshold value of approximately 6%. However, only PPVVM correlated with CI changes induced by volume expansion.
Under open-chest conditions, the effects of cyclic changes in intrathoracic pressure on cardiovascular structures are diminished because changes in pleural pressure have less pronounced effects on the right atrium. A previous study showed that PPV converges to a smaller value (<8%) within a narrow range after the thorax is opened.14 Unlike closed-chest conditions, the narrow range of PPV related to open-thorax conditions makes it difficult to discriminate between responders and nonresponders.15
In this study, the Valsalva manoeuvre was used to restore the reduced heart–lung interaction in open-chest conditions. Use of the Valsalva manoeuvre increases the effects of alveolar pressure on thoracic cardiovascular structures and therefore rewidens the range of PPV. We hypothesised that the augmentation of PPV by the Valsalva manoeuvre would be more prominent in hypovolaemic patients. Two recent studies have shown that PPV augmented by forced inspiration or the Valsalva manoeuvre successfully overcomes limitations according to reduced variation in pleural and transpulmonary pressures in spontaneously breathing patients.18,23 PPV calculated during the Valsalva manoeuvre in spontaneously breathing patients successfully predicted fluid responsiveness after volume expansion18 (area under ROC curve 0.98, 95% CI 0.84 to 0.99) with a threshold value of 52%, which is very similar to the one calculated in this study (55%). However, the typical four-phase arterial pressure response to the Valsalva manoeuvre involving systemic pressure overshoot seen in spontaneously breathing and conscious patients was not prominent in our study.18,24 This might have resulted from our study setting, which was conducted in mechanically ventilated patients under general anaesthesia.
Unexpectedly, baseline PPV also showed significant predictability for fluid responsiveness in the present study. This finding appeared to contradict results of recent studies in which PPV failed to predict fluid responsiveness in open-chest conditions.1,12,13 However, there are conflicting results regarding the usefulness of PPV as a predictor of fluid responsiveness in open-chest conditions,1,12,13,15–17 although the predictability of PPV for fluid responsiveness is theoretically weakened after the thorax is opened. There has been only one study using ROC analysis, which suggested the unreliability of PPV as a predictor of fluid responsiveness in open-chest conditions.12 However, in our opinion, that study was underpowered because there were only three nonresponders. In addition, two previous studies showed that PPV correlated with the fluid loading-related increase in CI16 and with global end-diastolic volume index15 in open-chest conditions.
In the present study, although there was no correlation between baseline PPV and volume-induced change in CI, PPV predicted fluid responsiveness in the ROC analysis. However, the predictability, sensitivity and specificity were all lower than those of PPVVM. Likelihood ratios, which indicate the diagnostic value of a test,25 also suggested that PPVVM was superior to conventional PPVs, although all three PPVs seemed to have a diagnostic value. Specifically, the lowest negative likelihood ratio of PPVVM together with the highest sensitivity and highest negative predictive value indicated that PPVVM is more sensitive than conventional PPVs for diagnosing fluid responsiveness under open chest conditions. Moreover, as we hypothesised that a test is significant when more than a 0.3 difference from null hypothesis exists in the area under ROC curve, only PPVVM was predictive of fluid responsiveness in our study. Moreover, considering both open chest-related convergence (<8%)14 and the possible existence of a grey zone26 of PPV, conventional PPV needs to be used carefully for fluid management in these conditions.
There are some limitations of this study. First, we assessed the predictability of PPVVM for fluid responsiveness under open-thorax but closed-pericardium conditions. Therefore, the usefulness of PPVVM under open-chest and open-pericardium conditions was not clarified. As most of the OPCAB surgery is performed with open pericardium, which might be left open after surgery, the clinical applicability of PPVVM could be significantly enhanced if it was assessed in the open-thorax and open-pericardium conditions. However, to our knowledge, this is the first study using adequately powered ROC curve analysis to identify reliable predictors of fluid responsiveness in open-chest conditions. Further studies will be needed to assess the predictability of PPVVM for fluid responsiveness under open-thorax and open-pericardium conditions because guides for adequate fluid management would be helpful during or after main procedures performed with the pericardium open in clinical practice. Second, our study sample consisted only of patients with preserved left ventricular function to obtain more homogeneous data and for clearer interpretations of the results. Therefore, the results of this study do not demonstrate the usefulness of PPVVM in patients with reduced left ventricular function. Patients with heart failure with marginal functional reserve are more vulnerable to volume loading and therefore require more delicate use of intravenous fluids. Within normal ranges, CI slightly decreased after volume expansion in fluid nonresponder group in this study. Therefore, further studies using more precise methods are required for patients with ventricular dysfunction. Finally, the Valsalva manoeuvre performed in our study has potentially deleterious haemodynamic effects in patients undergoing OPCAB. Our manoeuvre can cause a profound decrease in blood pressure in these patients and applying high inspiratory pressures might also not be desirable immediately after coronary anastomoses. Therefore, the Valsalva manoeuvre should be used with caution in these patients. In addition, more sophisticated methods of PPV augmentation should be investigated in future studies. However, no similar open-thorax studies have been conducted, and thus, we maintained 30 cmH2O pressure for 10 s, which was used successfully to predict fluid responsiveness in spontaneously breathing patients.18
In conclusion, we found that augmented PPV using a Valsalva manoeuvre is a clinically reliable predictor of fluid responsiveness in open-chest conditions. Therefore, the use of such augmented PPV would be helpful for fluid management in patients undergoing OPCAB.
Acknowledgements relating to this article
Assistance with the study: none
Financial support and sponsorship: none
Conflicts of interest: none
Presentation: this study was presented as a poster at the ASA 2015 annual meeting, 24 to 28 October 2015, in San Diego, California, USA.
1. Rex S, Schalte G, Schroth S, et al. Limitations of arterial pulse pressure variation and left ventricular stroke volume variation in estimating cardiac preload during open heart surgery. Acta Anaesthesiol Scand
2. Michard F. Changes in arterial pressure during mechanical ventilation. Anesthesiology
3. Michard F, Teboul JL. Using heart-lung interactions to assess fluid responsiveness during mechanical ventilation. Crit Care
4. Michard F, Boussat S, Chemla D, et al. Relation between respiratory changes in arterial pulse pressure and fluid responsiveness in septic patients with acute circulatory failure. Am J Respir Crit Care Med
5. Berkenstadt H, Margalit N, Hadani M, et al. Stroke volume variation as a predictor of fluid responsiveness in patients undergoing brain surgery. Anesth Analg
6. Reuter DA, Kirchner A, Felbinger TW, et al. Usefulness of left ventricular stroke volume variation to assess fluid responsiveness in patients with reduced cardiac function. Crit Care Med
7. Rex S, Brose S, Metzelder S, et al. Prediction of fluid responsiveness in patients during cardiac surgery. Br J Anaesth
8. Kramer A, Zygun D, Hawes H, et al. Pulse pressure variation predicts fluid responsiveness following coronary artery bypass surgery. Chest
9. Preisman S, Kogan S, Berkenstadt H, et al. Predicting fluid responsiveness in patients undergoing cardiac surgery: functional haemodynamic parameters including the Respiratory Systolic Variation Test and static preload indicators. Br J Anaesth
10. Hofer CK, Muller SM, Furrer L, et al. Stroke volume and pulse pressure variation for prediction of fluid responsiveness in patients undergoing off-pump coronary artery bypass grafting. Chest
11. Lee JH, Kim JT, Yoon SZ, et al. Evaluation of corrected flow time in oesophageal Doppler as a predictor of fluid responsiveness. Br J Anaesth
12. de Waal EE, Rex S, Kruitwagen CL, et al. Dynamic preload indicators fail to predict fluid responsiveness in open-chest conditions. Crit Care Med
13. Wyffels PA, Sergeant P, Wouters PF. The value of pulse pressure and stroke volume variation as predictors of fluid responsiveness during open chest surgery. Anaesthesia
14. De Blasi RA, Palmisani S, Cigognetti L, et al. Effects of sternotomy on heart-lung interaction in patients undergoing cardiac surgery receiving pressure-controlled mechanical ventilation. Acta Anaesthesiol Scand
15. Reuter DA, Goresch T, Goepfert MS, et al. Effects of mid-line thoracotomy on the interaction between mechanical ventilation and cardiac filling during cardiac surgery. Br J Anaesth
16. Reuter DA, Goepfert MS, Goresch T, et al. Assessing fluid responsiveness during open chest conditions. Br J Anaesth
17. Sander M, Spies CD, Berger K, et al. Prediction of volume response under open-chest conditions during coronary artery bypass surgery. Crit Care
18. Monge Garcia MI, Gil Cano A, Diaz Monrove JC. Arterial pressure changes during the Valsalva manoeuvre to predict fluid responsiveness in spontaneously breathing patients. Intensive Care Med
19. Pinsky MR. Using ventilation-induced aortic pressure and flow variation to diagnose preload responsiveness. Intensive Care Med
20. De Backer D, Heenen S, Piagnerelli M, et al. Pulse pressure variations to predict fluid responsiveness: influence of tidal volume. Intensive Care Med
21. Cannesson M, Slieker J, Desebbe O, et al. The ability of a novel algorithm for automatic estimation of the respiratory variations in arterial pulse pressure to monitor fluid responsiveness in the operating room. Anesth Analg
22. Aboy M, McNames J, Thong T, et al. A novel algorithm to estimate the pulse pressure variation index deltaPP. IEEE Trans Biomed Eng
23. Hong DM, Lee JM, Seo JH, et al. Pulse pressure variation to predict fluid responsiveness in spontaneously breathing patients: tidal vs. forced inspiratory breathing. Anaesthesia
24. De Lee GJ, Matthews MB, Sharpey-Schafer EP. The effect of the Valsalva manoeuver on the systemic and pulmonary arterial pressure in man. Br Heart J
25. Florkowski CM. Sensitivity, specificity, receiver-operating characteristic (ROC) curves and likelihood ratios: communicating the performance of diagnostic tests. Clin Biochem Rev
2008; 29 (Suppl 1):S83–S87.
26. Michard F, Chemla D, Teboul JL. Applicability of pulse pressure variation: how many shades of grey? Crit Care
* Both Jeong Jin Min and Tae-Kyong Kim contributed equally to this manuscript.© 2017 European Society of Anaesthesiology