In patients undergoing lung surgery, intra-operative fluid loading is frequently restricted to prevent pulmonary oedema which is the critical complication after lung surgery. However, during the peri-operative period, it is also essential to maintain optimal organ perfusion by appropriate fluid infusion. Therefore, to achieve the balance between preventing fluid overload and optimising organ perfusion,1 the practical index to guide fluid management, which can predict whether fluid loading will improve haemodynamic conditions in an individual patient,2 would be very valuable during lung surgery.
Although cardiac filling pressures [i.e. central venous pressure (CVP) and pulmonary arterial occlusion pressure) are generally used for fluid management, the value of these static preload indices has been questioned both because filling pressures cannot reflect ventricular filling volume3 and they fail to predict fluid responsiveness.4–6 Moreover, respiratory variations of arterial pressure [i.e. pulse pressure variation (PPV) and systolic pressure variation] can predict fluid responsiveness in mechanically ventilated patients under various conditions.4,5,7–13 Therefore, pressure variations are increasingly being advocated for fluid management. However, dynamic preload indices such as PPV are based on the cyclic changes of stroke volume and intra-thoracic pressure induced by positive pressure ventilation.14 Therefore, all factors affecting these conditions could influence the ability of dynamic preload indices to predict fluid responsiveness.15–19
There are two kinds of ventilating method used during one-lung ventilation (OLV) for lung surgery: conventional and protective ventilation techniques. Traditionally, conventional ventilation similar to two-lung ventilation has been used to prevent hypoxia during OLV, but can lead to acute lung injury after lung resection.20,21 Moreover, more effective lung isolation using fiberoptic bronchoscopy and the use of anaesthetic agents with fewer or no detrimental effects on hypoxic pulmonary vasoconstriction make hypoxia less common during OLV.21 In addition, a textbook recently changed the strategy of OLV to the protective one.22 Consequently, the protective ventilation characterised by low tidal volume, low inspired oxygen fraction (FIO2) and application of positive end-expiratory pressure (PEEP) is recently being advocated.20,21
However, to our knowledge, dynamic preload indices such as PPV have never been studied as predictors of fluid responsiveness during OLV. During conventional OLV, the same tidal volume as two-lung ventilation may double tidal volume applied to the one ventilated lung which can affect the cyclic alterations in intra-thoracic pressure induced by mechanical ventilation. In addition, the shunt through the non-ventilated lung may affect the degree of PPV.
Therefore, we hypothesised that PPV could predict fluid responsiveness during OLV under thoracotomy. Thus, we evaluated whether PPV could predict fluid responsiveness during conventional and protective OLV in patients undergoing lung surgery by receiver operating characteristics (ROCs) analysis, respectively.
Ethical approval for this study (H-0805-018-242) was provided by the Seoul National University Hospital Institutional Review Board, Seoul, Korea (President Sang Goo Shim) on 21 July 2008.
After obtaining our institutional review board approval and informed consent, patients scheduled for elective lung surgery requiring OLV were enrolled in this study. The Consolidated Standards of Reporting Trials (CONSORT) guidelines were followed with respect to the reporting of this randomised and controlled study. Patients were randomised using an internet-based computer program (http://www.randomizer.org) either to group P (patients receiving protective OLV with tidal volume of 6 ml kg−1, FIO2 of 0.5 and PEEP of 5 cmH2O) or group C (patients receiving conventional OLV with tidal volume of 10 ml kg−1, FIO2 of 1.0 and no PEEP).20 A investigator (N.-S. G.), who did not know the study protocol, generated the allocation sequence and assigned all patients to one of these two groups. Patients with known cardiac disease (except controlled hypertension), pre-operative arrhythmia and contraindications to oesophageal Doppler monitoring probe insertion (i.e. oesophageal stent, carcinoma of the oesophagus or pharynx, previous oesophageal surgery, oesophageal stricture, oesophageal varices, pharyngeal pouch and severe coagulopathy) were excluded.
Anaesthesia and one-lung ventilation
After the patient arrived in the operating room, our routine monitoring including pulse oxymetry, three-lead ECG and non-invasive arterial pressure was applied. Induction of anaesthesia was performed with propofol (2–3 mg kg−1) and fentanyl (3–4 μg kg−1). Following loss of consciousness, neuro-muscular block was achieved with rocuronium (0.6 mg kg−1). After anaesthesia induction, a left-sided double-lumen endo-bronchial tube was inserted and the position was confirmed by fiberoptic bronchoscopy. A radial arterial cannula was inserted and a central venous catheter was placed through right internal jugular vein. All pressure transducers were zeroed at mid-axillary line to ambient pressure and initial pressures were recorded with the patient in the supine position. After changing the patient's position to lateral decubitus, all pressure transducers were re-positioned at the same value of initially measured pressures in the supine position. Anaesthesia was maintained with inhaled sevoflurane.
Following the initiation of OLV, patients were ventilated with tidal volume of 6 ml kg−1 ideal body weight, FIO2 of 0.5 and PEEP of 5 cmH2O in group P. In group C, OLV with tidal volume of 10 ml kg−1 ideal body weight, FIO2 of 1.0 and no PEEP was applied. The respiratory rate was adjusted to maintain end-tidal carbon dioxide at 30–40 mmHg.
A Hemosonic oesophageal Doppler probe (Arrow International, Everett, Washington, USA) was inserted into the oesophagus for cardiac output (CO) monitoring. A well trained investigator (N.-S. G.) performed all oesophageal Doppler monitoring measurements during the study. The correct position of the oesophageal Doppler monitoring probe was confirmed by continuously measuring descending thoracic aorta blood velocity (Doppler transducer) and aortic diameter (M-mode echo transducer). Cardiac index (CI) was calculated as CO/body surface area (BSA) and BSA was calculated using the Du Bois formula [BSA = body weight (kg)0.425 × body length (m)0.725 × 0.20247].
For the calculation of PPV, arterial and capnography waveforms were recorded simultaneously for offline analysis. After recording, pulse pressure (PP; defined as the difference between the SBP and the DBP of the previous beat) was measured on a beat-to-beat basis using Adobe Photoshop CS2 software (Adobe Systems Inc., San Jose, California, USA). Maximal PP (PPmax) and minimal PP (PPmin) values were determined over a single respiratory cycle. PPV was calculated as follows:
The measurements were repeated on three successive respiratory cycles and averaged for statistical analysis.
The study was started after finishing chest opening and collapsing one lung totally. During OLV, values of heart rate (HR), mean arterial pressure (MAP), CVP, PPV and CO were measured before (T0) and 12 min (T1) after fluid loading in both groups. At each time point, CO was measured after the oesophageal Doppler probe was re-positioned wherein the monitor showed both good Doppler signal and the largest aortic diameter. All measurements were achieved in a haemodynamic steady state with no vasoactive medication. When haemodynamic instability was developed during the study period, the patient was excluded from the analysis. Volume loading was achieved by using 6% hydroxyethyl starch solution (HES 130/0.4; Voluven; Fresenius Kabi, Stans, Switzerland) at 7 ml kg−1 ideal body weight. In addition, the values of end-tidal carbon dioxide, mean airway pressure and peak inspiratory pressure were checked.
Statistical analysis was performed using SPSS 12.0 software (SPSS Inc., Chicago, Illinois, USA) and MedCalc 220.127.116.11 software (MedCalc Inc., Mariakerke, Belgium). All haemodynamic data were analysed as continuous variables and are expressed as mean ± SD. Comparisons of haemodynamic variables before and after volume expansion were assessed using a paired t-test. Student's t-test was used for comparing haemodynamic variables between groups and those between the responders and the non-responders. The χ2 test was used when indicated. The correlation between changes in CI and initial haemodynamic variables was assessed using linear regression. Percentage differences in oesophageal Doppler-derived CIs before and after volume expansion were used as principal indicators of fluid responsiveness. Patients were classified as the responders to fluid loading when increases in CI were at least 15%. To test the abilities of CVP and PPV to predict fluid responsiveness, areas under the ROC curves of the responders [area under the curve (AUC) = 0.5: no better than chance, no prediction possible; AUC = 1.0: best possible prediction] were calculated and compared using the Hanley–McNeil test in each group. P < 0.05 was considered statistically significant.
In total, 55 patients were screened for the recruitment into this study between March 2008 and November 2008. Of these, five patients met our exclusion criteria (two had valvular heart disease, two had atrial fibrillation and one had known coagulopathy). In addition, one patient in group C did not complete our study protocol because of a significant bleeding-induced unstable haemodynamic state during the study (Fig. 1). Patient and surgical characteristics are described in Table 1. No complication occurred in relation to this study.
Eighteen patients were responders and seven were non-responders. HR and CVP increased significantly in responders and non-responders after fluid loading (Table 2). MAP and CI increased and PPV decreased in only the responders related to volume expansion (Table 2). PPV before fluid loading correlated with the changes in CI according to fluid loading, but CVP did not (Fig. 2). Moreover, only PPV before volume expansion was able to predict fluid responsiveness in ROC analysis (Table 3 and Fig. 3). The area under ROC curve for PPV was significantly larger than that for CVP (P = 0.02). The optimal threshold value given by ROC analysis was 5.8% for PPV with a sensitivity of 72% and a specificity of 100%.
Thirteen patients were responders and 11 patients were non-responders. After volume expansion, MAP and CVP increased significantly in both responders and non-responders (Table 2). CI increased and PPV decreased significantly in only the responders after fluid loading (Table 2). PPV and CVP before volume loading did not correlate with changes in CI (Fig. 2), and ROC analysis revealed that neither was able to predict fluid responsiveness with sufficient statistical power (Table 3 and Fig. 3). There was no significant difference between the areas under ROC curves for PPV and CVP.
In this study, ROC analysis showed that PPV could predict fluid responsiveness under protective OLV, but not under conventional OLV. In addition, the threshold value of PPV to predict fluid responsiveness during protective OLV was lower in patients receiving two-lung ventilation.4,5,8,10–13
Although many previous studies have suggested that PPV at least around 10% could predict fluid responsiveness in two-lung ventilation, on the physiologic basis, PPV relies on variations of blood flow caused by the cyclic changes in intra-thoracic pressure during mechanical ventilation.2,17 Therefore, both mechanical ventilation method and significant shunt amount through the non-ventilated lung can influence the predictive value of PPV for fluid responsivenes, regardless of the patient's preload state.
During OLV, if the same tidal volume is applied, the ventilated lung is exposed to double the tidal volume of two-lung ventilation. This could increase right ventricular afterload and exaggerate the cyclic variation in stroke volume14 and PPV. Moreover, in previous studies, large tidal volume can increase the value of PPV without changes in volume status.15,16 Therefore, increased PPV of non-responders (7.7 ± 3.0%) abolished the difference with PPV of responders (8.2 ± 3.5%) during conventional OLV. This could explain why PPV failed to predict fluid responsiveness during conventional OLV. On the contrary, the tidal volume during protective OLV may be similar to two-lung ventilation and PPVs were different between responders and non-responders (responders vs. non-responders; 7.6 ± 2.8% vs. 4.5 ± 1.1%, P < 0.05) (Table 1).
In this study, the PPV threshold value to detect fluid responsiveness during protective OLV was 5.8%, almost half that for two-lung ventilation.4,5,8,10–13 During OLV, there is a 20–30% shunt through the non-ventilated lung even with optimal management.22 This shunt amount does not contribute to the generation of PPV because there is no cyclic change of intra-thoracic pressure in the non-ventilated lung. Therefore, the shunt through the non-ventilated lung would decrease the value of PPV, irrespective of the patient's preload state, explaining the lower PPV threshold value in this study.
In the present study, two types of OLV were used for testing the ability of PPV as a predictor of fluid responsiveness. Despite attempts to change OLV strategies toward protective ventilation,22 conventional OLV technique is still used in some clinical centres. Therefore, we used both types of OLV.
There are several limitations in our study. First, we measured CO with oesophageal Doppler. Although thermodilution is considered as the clinical standard method to measure CO, CO measured by oesophageal Doppler correlated well with that measured by thermodilution.23,24 In addition, we determined CO when the oesophageal Doppler monitor showed the largest aortic diameter which improves oesophageal Doppler accuracy in assessing the haemodynamic effects of volume loading.25 Second, HR increased after fluid loading in group P which could affect the increase in CO. However, the ability of PPV to predict fluid responsiveness is not limited by HR alone because dynamic preload indices occur in hypovolemic conditions with HR of 70–150 beats min−1.10,26–28 Therefore, the slight increase in HR might not affect our results. Third, we did not measure lung compliance and intra-thoracic pressure in this study. Therefore, we could not explain exactly the physiologic difference between conventional and protective OLV. However, because we used an oesophageal Doppler system to measure CO, it is technically difficult to use two oesophageal probes simultaneously. Lastly, pressure-controlled ventilation was not used in patients receiving protective OLV, despite its clinical use. Because PPV is not able to predict fluid responsiveness during pressure support ventilation,29 PPV might also have failed to predict fluid responsiveness if pressure-controlled protective OLV was used here. However, this topic is beyond our study and pressure-controlled ventilation is not mandatory for protective OLV.
In conclusion, PPV can predict fluid responsiveness during protective OLV, but not conventional OLV. The threshold value of PPV as a predictor of fluid responsiveness during protective OLV was lower than that during two-lung ventilation. PPV could be useful to guide fluid management during OLV if protective ventilation is applied.
We would like to thank Dr Joo-Hyun Kim, Dr Young Tae Kim, Dr Chang-Hyun Kang and all the other members of the Department of Thoracic and Cardiaovascular Surgery in Seoul National University Hospital for their kind co-operation with our study. This work was only supported by the Department of Anesthesiology and Pain Medicine, Seoul National University Hospital, Seoul National College of Medicine, Seoul, Korea.
1 Rocca GD, Costa MG. Preload indexes in thoracic anesthesia. Curr Opin Anaesthesiol 2003; 16:69–73.
2 Michard F. Changes in arterial pressure during mechanical ventilation. Anesthesiology 2005; 103:419–428.
3 Kumar A, Anel R, Bunnell E, et al
. Pulmonary artery occlusion pressure and central venous pressure fail to predict ventricular filling volume, cardiac performance, or the response to volume infusion in normal subjects. Crit Care Med 2004; 32:691–699.
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 2000; 162:134–138.
5 Michard F, Teboul J. Predicting fluid responsiveness in ICU patients: a critical analysis of the evidence. Chest 2002; 121:2000–2008.
6 Osman D, Ridel C, Ray P, et al
. Cardiac filling pressures are not appropriate to predict hemodynamic response to volume challenge. Crit Care Med 2007; 35:64–68.
7 Berkenstadt H, Margalit N, Hadani M, et al
. Stroke volume variation as a predictor of fluid responsiveness in patients undergoing brain surgery. Anesth Analg 2001; 92:984–989.
8 Kramer A, Zygun D, Hawes H, et al
. Pulse pressure variation predicts fluid responsiveness following coronary artery bypass surgery. Chest 2004; 126:1563–1568.
9 Rex S, Brose S, Metzelder S, et al
. Prediction of fluid responsiveness in patients during cardiac surgery. Br J Anaesth 2004; 93:782–788.
10 Reuter DA, Goepfert MS, Goresch T, et al
. Assessing fluid responsiveness during open chest conditions. Br J Anaesth 2005; 94:318–323.
11 Wiesenack C, Fiegl C, Keyser A, et al
. Assessment of fluid responsiveness in mechanically ventilated cardiac surgical patients. Eur J Anaesthesiol 2005; 22:658–665.
12 Lee J, Kim J, Yoon SZ, et al
. Evaluation of corrected flow time in oesophageal Doppler as a predictor of fluid responsiveness. Br J Anaesth 2007; 99:343–348.
13 Huang C, Fu J, Hu H, et al
. Prediction of fluid responsiveness in acute respiratory distress syndrome patients ventilated with low tidal volume and high positive end-expiratory pressure. Crit Care Med 2008; 36:2810–2816.
14 Michard F, Teboul JL. Using heart-lung interactions to assess fluid responsiveness during mechanical ventilation. Crit Care 2000; 4:282–289.
15 Reuter D, Bayerlein J, Goepfert MS, et al
. Influence of tidal volume on left ventricular stroke volume variation measured by pulse contour analysis in mechanically ventilated patients. Intensive Care Med 2003; 29:476–480.
16 Oliveira R, Azevedo L, Park M, Schettino GP. Influence of ventilatory settings on static and functional haemodynamic parameters during experimental hypovolaemia. Eur J Anaesthesiol 2009; 26:66–72.
17 De Backer D, Taccone FS, Holsten R, et al
. Influence of respiratory rate on stroke volume variation in mechanically ventilated patients. Anesthesiology 2009; 110:1092–1097.
18 Rex S, Schälte 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 2007; 51:1258–1267.
19 de Waal EE, Rex S, Kruitwagen CL, et al
. Dynamic preload indicators fail to predict fluid responsiveness in open-chest conditions. Crit Care Med 2009; 37:510–515.
20 Sentürk M. New concepts of the management of one-lung ventilation. Curr Opin Anaesthesiol 2006; 19:1–4.
21 Lohser J. Evidence-based management of one-lung ventilation. Anesthesiol Clin 2008; 26:241–272.
22 Slinger PD, Campos JH. Anesthesia for thoracic surgery. In: Miller RD, editor. Miller's Anesthesia. 7th ed. Philadelphia: Elsevier Churchill Livingstone; 2009. pp. 1819–1887.
23 Baillard C, Cohen Y, Fosse JP, et al
. Haemodynamic measurements (continuous cardiac output
and systemic vascular resistance) in critically ill patients: transoesophageal Doppler versus continuous thermodilution. Anaesth Intensive Care 1999; 27:33–37.
24 Leone D, Servillo G, De Robertis E, et al
. Monitoring cardiac output
: esophageal doppler vs thermodilution. Minerva Anestesiol 1998; 64:351–356.
25 Monnet X, Chemla D, Osman D, et al
. Measuring aortic diameter improves accuracy of esophageal Doppler in assessing fluid responsiveness. Crit Care Med 2007; 35:477–482.
26 Feissel M, Michard F, Mangin I, et al
. Respiratory changes in aortic blood velocity as an indicator of fluid responsiveness in ventilated patients with septic shock. Chest 2001; 119:867–873.
27 Vieillard-Baron A, Chergui K, Rabiller A, et al
. Superior vena caval collapsibility as a gauge of volume status in ventilated septic patients. Intensive Care Med 2004; 30:1734–1739.
28 Preisman S, DiSegni E, Vered Z, Perel A. Left ventricular preload and function during graded haemorrhage and retranfusion in pigs: analysis of arterial pressure waveform and correlation with echocardiography. Br J Anaesth 2002; 88:716–718.
29 Perner A, Faber T. Stroke volume variation does not predict fluid responsiveness in patients with septic shock on pressure support ventilation. Acta Anaesthesiol Scand 2006; 50:1068–1073.