Secondary Logo

Journal Logo

Comparison of the PaO2/FiO2 ratio in sternotomy vs. thoracotomy in mitral valve repair: a randomised controlled trial

Kang, Woon-Seok; Yoon, Tae-Gyoon; Kim, Tae-Yop; Kim, Seong-Hyop

European Journal of Anaesthesiology: November 2011 - Volume 28 - Issue 11 - p 807–812
doi: 10.1097/EJA.0b013e32834ad993
Cardiac anaesthesia

Objective Cardiac surgery through a thoracotomy using one-lung ventilation (OLV) is thought to be associated with worse postoperative pulmonary gas exchange than sternotomy using two-lung ventilation (TLV), but this has not been confirmed yet. We, therefore, compared postoperative pulmonary gas exchange after mitral valve repair between sternotomy (group TLV) and thoracotomy (group OLV).

Design Randomised controlled study.

Setting University teaching hospital.

Participants Cardiac surgery patients.

Intervention Sternotomy or thoracotomy was used for mitral valve repair.

Measurements The ratio of arterial partial pressure of oxygen (PaO2) to fraction of inspired oxygen (FiO2) was compared in both groups before induction of anaesthesia (T0) and just before departure from the operating room to the ICU (T1). Fluid administration, transfusion requirements and urine output were checked intraoperatively. Postoperative haemoglobin (Hb), haematocrit (Hct) and creatinine were evaluated. Cardiopulmonary bypass (CPB) time, intubation time and ICU stay were also recorded.

Results The PaO2/FiO2 ratio (mean ± SD) at T1 was significantly lower than at T0 in both groups (326.9 ± 120.1 vs. 431.9 ± 73.7 mmHg in group TLV, P < 0.001; 374.9 ± 130.9 vs. 445.4 ± 73.7 mmHg in group OLV, P = 0.001), but did not differ significantly between the two groups. The doses of inotropes and vaopressors used were not significantly different between the groups. Intraoperative fluid administration, transfusion requirements, urine output and postoperative Hb/Hct and creatinine did not differ significantly between the groups. CPB time, intubation time and ICU stay also did not differ significantly between the groups.

Conclusion Perioperative pulmonary function following OLV via a thoracotomy was not significantly worse than that following TLV via a sternotomy in mitral valve repair.

Clinical trial registration Not registered.

From the Department of Anaesthesiology and Pain medicine, Konkuk University Hospital and Research Institute of Medical Science, Konkuk University School of Medicine, Seoul, South Korea

Correspondence to Seong-Hyop Kim, MD, PhD, Department of Anaesthesiology and Pain medicine, Konkuk University School of Medicine, Hwayang-dong, Gwangjin-gu, Seoul 143-729, South KoreaTel: +82 2 2030 5454; fax: +82 2 2030 5449; e-mail:

Published online 5 October 2011

Back to Top | Article Outline


Since Julian et al.1 introduced the sternotomy in the 1950s, it has become the approach of choice for cardiac surgical procedures, but more recently, evolving technologies have increased the use of minimally invasive cardiac approaches using computer-assisted or robotic-assisted procedures via a thoracotomy. After cardiac surgical procedures with cardiopulmonary bypass (CPB), impaired pulmonary gas exchange is a frequent occurrence.2,3 One-lung ventilation (OLV) may lead to a greater deterioration in lung function through ventilation/perfusion mismatch, shunt, ischaemia/reperfusion injury and the inflammatory process from re-expansion of the collapsed lung.4–6 For these reasons, cardiac surgical procedures through a thoracotomy (OLV) are thought to be associated with greater postoperative pulmonary gas exchange impairment than sternotomy (two-lung ventilation, TLV), but this has not been proven yet.

We hypothesised that pulmonary function after thoracotomy (OLV) would be worse than after sternotomy (TLV) in mitral valve repair. We compared postoperative pulmonary gas exchange after mitral valve repair between a sternotomy (group TLV) and a thoracotomy (group OLV) in this randomised controlled trial.

Back to Top | Article Outline


Study population

Approval for this prospective study was granted by the Institutional Review Board of Konkuk University Medical Center, Seoul, South Korea (KUH 1160016, 23 November 2010). Informed consent was obtained from patients undergoing mitral valve repair in a university teaching hospital between November 2010 and March 2011. Exclusion criteria were as follows: urgent or emergency case; other concurrent valvular surgery; patient age less than 16 years; reduced left and right ventricular function (ejection fraction < 40%); previous respiratory disease; arterial partial pressure of oxygen (PaO2)/fraction of inspired oxygen (FiO2) ratio less than 300 mmHg before anaesthesia; preoperative dysrhythmia; intracardiac shunt; severe renal disease; severe hepatic disease; haematocrit (Hct) less than 30% before anaesthesia; or re-operation for cardiac valvular disease. The trial design of the study was parallel and the patients were randomly allocated before anaesthesia (allocation ratio = 1 : 1) using sealed envelopes to undergo either TLV or OLV. Cardiac anaesthesiologists, cardiac surgeons, nurses and perfusionists were blind to the study. All data were collected by trained observers who did not participate in patient care and who were blinded to the study.

Back to Top | Article Outline

Anaesthetic regimen

After establishing routine invasive arterial blood pressure and non-invasive patient monitoring (pulse oximetry, ECG, bispectral index and cerebral oximetry), anaesthesia was induced and maintained using target-controlled infusion (Orchestra Base Primea, Fresenius Vial, Brezins, France) of propofol (target concentration 1.2–1.3 μg ml−1) and remifentanil (target concentration 10–20 ng ml−1). Muscle relaxation was obtained with a bolus of rocuronium while monitoring peripheral neuromuscular transmission. A pulmonary artery catheter and a transoesophageal echocardiography for perioperative cardiac monitoring were inserted after induction.

Patients in group TLV were intubated with an endotracheal tube (Hi-Lo, Mallinckrodt Medical, Athlone, Ireland) and the following ventilator (ADU, Datex-Ohmeda, Bromma, Sweden) settings were used: 4 l min−1, consisting of air (3 l min−1) and oxygen (1 l min−1), tidal volume = ideal body weight × 7 ml. Respiratory rate was controlled using end-tidal carbon dioxide pressure (EtCO2) (S/5 Compact Anaesthesia Monitor, Datex-Ohmeda), which was kept between 35 and 40 mmHg without positive end-expiratory pressure (PEEP), and the inspiratory/expiratory ratio was set to 1 : 2. Ventilation was stopped at aortic cross-clamping (ACC) and restarted at release of ACC.

Patients in group OLV were intubated with a univent tube (Phycon, Fuji Systems, Tokyo, Japan). The appropriate position and volume of the bronchial balloon for OLV were confirmed by fiberoptic bronchoscopy (LF-GP, Olympus Medical Systems, Tokyo, Japan) before and after the patients were placed in the lateral position. OLV was started just before the right fourth intercostal space was spilt and ended after thorax closure. The ventilator settings during OLV were the same as in group TLV, except the total fresh gas flow of 4 l min−1 consisted of 100% oxygen. At the start of CPB and ACC, the bronchial balloon of the univent tube was released and OLV was stopped. After ACC release, OLV was resumed. Before the conversion from OLV to TLV, one breath holding with TLV and 25 cmH2O for 5 s was applied. The procedure was performed through a thoracotomy with the back elevated 30° to the right. Those patients in group OLV who required TLV because oxygen saturation was below 90% despite PEEP to the dependent (left) lung, continuous positive airway pressure (CPAP) to the non-dependent (right) lung or peak inspiratory pressure (PIP) of more than 30 cmH2O during OLV, were not included in the analysis. The univent tube was not exchanged in the operating room.

Systemic mean arterial blood pressure above 60 mmHg was maintained during anaesthesia. Cardiac index was maintained above 2.0 l min−1 m−2 during anaesthesia except during CPB.

Back to Top | Article Outline

Cardiopulmonary bypass regimen

After administering heparin 300 units kg−1, the arterial and venous cannulation for CPB was performed with an activated clotting time above 450 s. In group TLV, the ascending aorta was used for arterial cannulation and the superior and inferior vena cavae for venous cannulation. In group OLV, the right femoral artery was used for arterial cannulation and the superior vena cava and femoral vein for venous cannulation. Activated clotting time was maintained above 450 s during CPB which was conducted using a reservoir, membrane oxygenator, roller pump and heat exchanger. The priming volume for CPB consisted of isotonic saline, 20% mannitol, NaHCO3, 20% albumin, heparin, antibiotics and calcium gluconate. Steroids were not administered intraoperatively. The flow of CPB was initiated at 60 ml kg−1 min−1 and was adjusted by haemodilution and temperature. Antegrade (via coronary ostium) or retrograde (via coronary sinus) cold blood cardioplegic solution (20 ml kg−1) was used for cardiac protection after ACC. Blood cardioplegic solution (1 l) contained the following components: NaCl 6.43 g, KCl 1.193 g, CaCl2 0.176 g and MgCl2 3.253 g (pH 7.4, 4–8°C).

Back to Top | Article Outline


The FiO2 and PaO2 were measured before induction (T0) and just before departure from the operating room to the ICU (T1) and the PaO2/FiO2 ratio was calculated. Crystalloid solution was administered according to maintenance fluid requirements, redistribution and evaporative surgical fluid losses, by body weight (4 ml kg−1) and colloid solution was administered to replace the surgical blood loss until the laboratory values reached the threshold for transfusion. Red blood cells were transfused when the Hct was below 20% during and 30% after CPB. Fresh frozen plasma, platelet concentrate and cryoprecipitate were not transfused during the surgical procedure in the operating room.

Fluid administration, transfusion requirements and urine output were checked intraoperatively. Postoperative haemoglobin (Hb), Hct and creatinine were evaluated at T1. CPB time, intubation time and ICU stay were also recorded.

Back to Top | Article Outline

Protocols for extubation and ICU discharge

The extubation regimen was started when the following criteria were met7: haemodynamically stable, urine output of at least 0.5 ml kg−1 h−1, temperature more than 36°C and chest tube drainage less than 100 ml h−1. At that time (45 min before anticipated extubation), all patients received morphine 0.2 mg kg−1. Vital signs, pain and sedation were assessed and an additional bolus of morphine 5 mg was administered according to clinical needs.

Patients were extubated when the following criteria were met: adequate response to commands, adequate respiratory efforts, oxygen saturation measured by pulse oximetry of at least 95% at a FiO2 of 0.5 or less, pH of at least 7.3 and an arterial partial pressure of carbon dioxide 55 mmHg or less.

The patients were eligible for transfer from the ICU when the following criteria were met: adequate cardiac stability with no haemodynamically significant arrhythmia, pulse oximetry of at least 90% at a FiO2 of 0.5 or less by a face mask, urine output of more than 0.5 ml kg−1 h−1, chest tube drainage less than 50 ml h−1, no intravenous inotrope or vasopressor therapy and no seizure activity.

Back to Top | Article Outline

Statistical analysis

The primary outcome variable was the PaO2/FiO2 ratio at T1. A PaO2/FiO2 ratio at T1 of 329.3 ± 111.9 mmHg was calculated from 10 previous patients undergoing TLV. For the PaO2/FiO2 ratio at T1, a minimum detected difference of 20% between the groups was considered clinically significant. A sample size of 47 in each group was calculated to be appropriate to achieve a power of 0.8 and an α value of 0.05. Secondary outcome variables were intubation time and ICU stay. By applying the same method of sample size determined for intubation time (744.1 ± 218.4 min) and ICU stay (3471.7 ± 1219.0 min), sample sizes of 32 and 50, respectively, in the respective groups were calculated. Statistical analyses were conducted using the SigmaStat software (version 3.1, SYSTAT Software, San Jose, California, USA). Continuous variables were analysed using the t-test between two groups and the paired t-test within each group and categorical variables using the χ2-test. Data are expressed as mean ± SD and number. A P value less than 0.05 was considered significant.

Back to Top | Article Outline


During the study, 198 cardiac surgical procedures were performed and 125 patients were eligible for the study. Of these 20 were excluded: 10 for other concurrent valvular surgery; three for previous respiratory disease; five for preoperative dysrhythmia; and two for re-operation for cardiac valvular disease. Fifty-two patients in group TLV and 53 patients in group OLV entered the study. However, two in group TLV were excluded because it was necessary to re-establish CPB due to surgical bleeding after administering protamine, and three patients in group OLV did not tolerate OLV, leaving 50 patients in each group for the final analysis (Fig. 1). The study was ended when the planned sample size (50 patients for each group) was achieved and further evaluation or follow-up after ICU discharge was not required. There were no adverse events in the study groups.

Fig. 1

Fig. 1

The patients’ characteristics were similar between the groups (Table 1). The duration of OLV in group OLV was 87 ± 37 min. In both groups, the PaO2/FiO2 ratio at T1 was significantly lower than at T0: 431.9 ± 73.7 mmHg at T0 vs. 326.9 ± 120.1 mmHg at T1 in Group TLV, P < 0.001 and 445.4 ± 73.7 mmHg at T0 vs. 374.9 ± 130.9 mmHg at T1 in group OLV (P = 0.001) (Table 2). In group OLV, PEEP to the dependent (left) lung was applied independently in 15 patients, together with CPAP to the non-dependent (right) lung in two patients. There were no patients with PIP of at least 30 cmH2O during OLV.

Table 1

Table 1

Table 2

Table 2

The difference in the PaO2/FiO2 ratio between the two groups was not significant (Table 2 and Fig. 2). Inotrope and vasopressor doses, intraoperative fluid administration, transfusion requirements and urine output were not significantly different between the groups. Postoperative Hb/Hct and creatinine also did not differ significantly between the groups. Intubation time and ICU stay were similar in the groups [group TLV vs. OLV: 672.5 ± 239.9 vs. 712.1 ± 239.8 min intubation time (P = 0.44); 2798.0 ± 790.4 vs. 2784.4 ± 985.7 min ICU stay (P = 0.96)] (Table 2).

Fig. 2

Fig. 2

Back to Top | Article Outline


In this study, immediate postoperative pulmonary gas exchange was decreased compared with the preoperative state regardless of TLV or OLV in mitral valve repair, but there was no significant difference between the groups with respect to this or other observations made.

Tribble et al.8 reported that postoperative pulmonary function, based on a comparison of mechanical ventilation duration, did not differ significantly between anterolateral thoracotomy and median sternotomy. However, they did not measure variables of pulmonary function, such as laboratory data, and did not come to a clear conclusion regarding the effects of OLV on postoperative pulmonary function; their interest was in re-operation for cardiac valvular surgery. Braxton et al.9 suggested that a PaO2/FiO2 ratio less than 300 mmHg was a prognostic indicator for an intrapulmonary shunt of approximately 20% and was associated with prolonged ICU and hospital stays. Thus, we used it to detect pulmonary dysfunction in our study. All patients who underwent mitral valve repair were admitted to the ICU where TLV regardless of sternotomy or thoracotomy was performed, except in special situations. As cardiopulmonary function improved in ICU, the PaO2/FiO2 ratio rose in line with the duration of stay. Therefore, we regarded the nadir of the PaO2/FiO2 ratio to be just before departure from the operating room to the ICU.

There may be several reasons why we failed to find a difference between the two groups in the PaO2/FiO2 ratio at T1. First, lung injury, as shown by biochemical and histopathological markers in animals, increased with the duration of OLV.10 This might affect the amounts of oxygen-free radicals generated.5 More than 1-h OLV might generate severe oxidative stress due to lung re-expansion, leading to cardiovascular complications. More prolonged OLV (> 90 min) can lead to clinically significant complications, such as acute respiratory failure, pulmonary hypertension and cardiac arrhythmia.6 In our study, the OLV duration was 87 ± 37 min, including the CPB time, but from the start of CPB to ACC and from ACC release to CPB weaning, pulmonary blood flow would have been much less than during the other periods of OLV. That is, the practical OLV duration may be shorter than that measured and factors that influence the deleterious effects on pulmonary gas exchange during OLV would be limited. Second, van der Werff et al.11 reported that patients with PIP of at least 40 cmH2O were prone to develop radiological signs of lung oedema. The mechanisms of lung injury related to a high PIP were increased microvascular–alveolar permeability due to stretch-activated cation channels, oxygen-derived free radicals, activated neutrophils and the upregulation of cytokines in the lung.12 In our study, a PIP of at least 30 cmH2O in OLV was an exclusion criterion and no patients met this requirement, limiting the stretch injury resulting from overpressure of the lung. Third, perioperative fluid administration less than 2 l has little effect on the postoperative pulmonary function,4,13 although there is no clear evidence that fluid restriction prevents acute lung injury.14 In our study, the total intraoperative fluid administration did not differ between the two groups (1917.6 ± 760.9 ml in group TLV and 1851.2 ± 678.0 ml in group OLV, P = 0.74). Fourth, the transfusion requirements and postoperative Hb/Hct did not differ between the two groups. This indicated that any adverse effect on pulmonary function related to transfusion was of the same extent in both groups. Finally, although the effect is small, total intravenous anaesthesia might lessen ventilation/perfusion mismatch and pulmonary shunt when compared with inhalational anaesthesia.15

To compare postoperative pulmonary function between TLV and OLV in mitral valve repair, time course evaluations are needed, but we chose to evaluate only the immediate postoperative PaO2/FiO2 ratio in an attempt to exclude bias due to postoperative pain, transfusion reaction and the state of sedation.

The different FiO2 (0.4 in TLV vs. 1.0 in OLV) might be associated with no difference in the postoperative PaO2/FiO2 ratio between sternotomy and thoracotomy. Oxygenation generally increases as FiO2 increases, but when FiO2 rises above 0.5 there is a contradictory decrease in oxygenation because of absorption atelectasis.16,17 In several animal studies, high FiO2 was found to induce pulmonary inflammation, histological changes and eventually pulmonary dysfunction.18,19 These findings, however, resulted from exposure to oxygen over several hours. In this study, because the duration of OLV was less than 90 min, the effect of high FiO2 was limited. After weaning from CPB, many factors such as inflammatory reaction from CPB, atelectasis of the dependent lung and compression of the lung by surgical manipulation can frequently result in hypoxemia (SpO2 < 90%), and we should manage OLV to prevent and treat hypoxemia.20–22 OLV with FiO2 1.0 is usually used in clinical situations, so we used TLV with FiO2 0.4 and OLV with FiO2 1.0 in this study.

Changes in haemodynamic variables also can influence oxygenation. However, the doses of inotropes and vasopressors were similar, limiting the influence of haemodynamic changes in the two groups.

Two further limitations of our study should be considered. First, the patients with a PaO2/FiO2 ratio of less than 300 mmHg before anaesthesia induction were excluded, ensuring that those in the study were healthy. Differences in postoperative pulmonary function between TLV and OLV are less likely to be found with healthy patients. However, the investigative process should begin with healthy groups and proceed to the less healthy. Second, three patients in group OLV [5.6% (3/53)] did not tolerate OLV during the surgical procedure. Intermittent TLV during OLV is not complete or pure OLV, and accordingly the patients were excluded. The conversion to TLV was because of hypoxia during OLV and was a necessary step for the safety of the patients. If these patients had been included in the analysis, the immediate postoperative PaO2/FiO2 ratio in group OLV would have been better than the values reported.

In conclusion, OLV via thoracotomy did not result in adverse perioperative pulmonary function, compared with TLV via sternotomy, in mitral valve repair.

Back to Top | Article Outline


No assistance, financial support or sponsorship was received for the completion of this study.

The English in this document has been checked by at least two professional editors, both native speakers of English. For a certificate, please see

Back to Top | Article Outline


1. Julian OC, Lopez-Belio M, Dye WS, et al. The median sternal incision in intracardiac surgery with extracorporeal circulation: a general evaluation of its use in heart surgery. Surgery 1957; 42:753–761.
2. Lohser J. Evidence-based management of one-lung ventilation. Anesthesiol Clin 2008; 26:241–272.
3. Ng CS, Wan S, Yim AP, Arifi AA. Pulmonary dysfunction after cardiac surgery. Chest 2002; 121:1269–1277.
4. Grichnik KP, Clark JA. Pathophysiology and management of one-lung ventilation. Thorac Surg Clin 2005; 15:85–103.
5. Misthos P, Katsaragakis S, Milingos N, et al. Postresectional pulmonary oxidative stress in lung cancer patients. The role of one-lung ventilation. Eur J Cardiothorac Surg 2005; 27:379–382.
6. Misthos P, Katsaragakis S, Theodorou D, et al. The degree of oxidative stress is associated with major adverse effects after lung resection: a prospective study. Eur J Cardiothorac Surg 2006; 29:591–595.
7. De Hert SG, Van der Linden PJ, Cromheecke S, et al. Choice of primary anesthetic regimen can influence intensive care unit length of stay after coronary surgery with cardiopulmonary bypass. Anesthesiology 2004; 101:9–20.
8. Tribble CG, Killinger WA Jr, Harman PK, et al. Anterolateral thoracotomy as an alternative to repeat sternotomy for replacement of the mitral valve. Ann Thorac Surg 1987; 43:380–382.
9. Braxton JH, Higgins RS, Schwann TA, et al. Reoperative mitral valve surgery via right thoracotomy: decreased blood loss and improved hemodynamics. J Heart Valve Dis 1996; 5:169–173.
10. Tekinbas C, Ulusoy H, Yulug E, et al. One-lung ventilation: for how long? J Thorac Cardiovasc Surg 2007; 134:405–410.
11. van der Werff YD, van der Houwen HK, Heijmans PJ, et al. Postpneumonectomy pulmonary edema: a retrospective analysis of incidence and possible risk factors. Chest 1997; 111:1278–1284.
12. Dos Santos CC, Slutsky AS. Invited review: mechanism of ventilator-induced lung injury – a perspective. J Appl Physiol 2000; 89:1645–1655.
13. Slinger PD. Perioperative fluid management for thoracic surgery: the puzzle of postpneumonectomy pulmonary edema. J Cardiothorac Vasc Anesth 1995; 9:442–451.
14. Sentürk M. New concepts of the management of one-lung ventilation. Curr Opin Anaesthesiol 2006; 19:1–4.
15. Beck DH, Doepfmer UR, Sinemus C, et al. Effects of sevoflurane and propofol on pulmonary shunt fraction during one-lung ventilation for thoracic surgery. Br J Anaesth 2001; 86:38–43.
16. Rothen HU, Sporre B, Engberg G, et al. Influence of gas composition on recurrence of atelectasis after a reexpansion maneuver during general anesthesia. Anesthesiology 1995; 82:832–842.
17. Miller RD. Respiratory physiology. In: Hedenstierna G, editor. Miller's anesthesia. 7th ed. Philadelphia, Pennsylvania, USA: Churchill Livingstone Elsevier; 2010. pp. 361–392.
18. Fracica PJ, Knapp MJ, Piantadosi CA, et al. Responses of baboons to prolonged hyperoxia: physiology and qualitative pathology. J Appl Physiol 1991; 71:2352–2362.
19. Szarek JL, Ramsay HL, Andringa A, Miller ML. Time course of airway hyperresponsiveness and remodeling induced by hyperoxia in rats. Am J Physiol 1995; 269:L227–L233.
20. Miller RD. Anesthesia for thoracic surgery. In: Slinger PD, Compos JH, editors. Miller's anesthesia. 7th ed. Philadelphia, Pennsylvania, USA: Churchill Livingstone Elsevier; 2010. pp. 1819–1888.
21. Benumof J. Conventional and differential lung management of one-lung ventilation. In: Benumof J, editor. Anaesthesia for thoracic surgery. 1st ed. Philadelphia, PA, USA: WB Saunders; 1987. pp. 271–284.
22. Yang HJ, Kim DOk, Choi YK, et al. The arterial oxygenation effects according to different ventilatory modes during one lung ventilation. Korean J Anesthesiol 1995; 29:414–422.

cardiac surgical procedure; pulmonary gas exchange; sternotomy; thoracotomy

© 2011 European Society of Anaesthesiology