Respiratory failure after cardiopulmonary bypass (CPB) remains a major complication after cardiac surgery. The authors tested the hypothesis that atelectasis is an important factor responsible for the increase in intrapulmonary shunt after CPB.
Six pigs received standard CPB (bypass group). Six other pigs had the same surgery but without CPB (sternotomy group). Another six pigs were anesthetized for the same duration but without any surgery (control group). The ventilation-perfusion distribution was measured with the inert gases technique, extravascular lung water was quantified by the double-indicator distribution technique, and atelectasis was analyzed by computed tomography.
Intrapulmonary shunt increased markedly after bypass but was unchanged over time in the control group (17.9 +/- 6.2% vs. 3.5 +/- 1.2%; P < 0.0001). Shunt also increased in the sternotomy group (10 +/- 2.6%; P < 0.01 compared with baseline) but was significantly lower than in the bypass group (P < 0.01). Extravascular lung water was not significantly altered in any group. The pigs in the bypass group showed extensive atelectasis (32.3 +/- 28.7%), which was significantly larger than in the two other groups. The pigs in the sternotomy group showed less atelectasis (4.1 +/- 1.9%) but still more (P < 0.05) than the controls (1.1 +/- 1.6%). There was good correlation between shunt and atelectasis when all data were pooled (R2 = 0.67; P < 0.0001).
Atelectasis is produced to a much larger extent after CPB than after anesthesia alone or with sternotomy and it explains most of the marked post-CPB increase in shunt and hypoxemia. Surgery per se contributes to a lesser extent to postoperative atelectasis and gas exchange impairment.
Pulmonary dysfunction is a well-documented complication of cardiopulmonary bypass (CPB) since the earliest days of cardiac surgery. [1,2] Despite improvements in surgical techniques, extracorporeal oxygenation equipment, and anesthesia management, postoperative pulmonary complications remains a major problem. [3,4] The incidence varies, from 2%, if defined as adult respiratory distress syndrome,  to 64%, if defined as atelectasis apparent on conventional chest roentgenograms.  Hammermeister et al.  reported an 8% incidence of pulmonary complications in a group of 10,634 patients. Furthermore, patients with pulmonary complications had a mortality rate of 25%.
The pulmonary complication, called “postperfusion lung syndrome” in the early era of cardiac surgery,  has stimulated a many studies in humans and in animals. Many possible causes have been considered, including intravascular microaggregates or blood cell damage ; leukocyte activation [9,10]; interstitial lung edema,  which can be aggravated by volume overload [12,13]; lung tissue hypoxia ; and alterations in surfactant activity. 
Alveolar collapse also has been proposed as an important cause of postoperative respiratory dysfunction.  Using computed tomography (CT), researchers found that general anesthesia induces atelectasis in nearly all patients, and this is correlated to intrapulmonary shunt. [16–18] It may be assumed that atelectasis also can be found after cardiac surgery, especially as the lungs are often open to the atmosphere under cardiopulmonary bypass, which facilitates lung collapse. If this is so, it may contribute to the high incidence of intrapulmonary shunting, and sometimes severe hypoxemia, that is frequently seen in persons having cardiac surgery. [8,19–23] Hachenberg et al.  found a 26.4% incidence of shunt in 11 patients with respiratory dysfunction in the early postoperative period, and this was well correlated with atelectasis as seen on CT scanning. Recently, the same group found that the increase in venous admixture is caused by “true shunt” and very little by perfusion of poorly ventilated lung regions, as assessed by multiple inert gas elimination.  To what extent such atelectasis remains or develops regularly after CPB is not clear.
Our aim in this study was to evaluate the amount of lung collapse and intrapulmonary shunt that arises after extracorporeal oxygenation in a pig model.
Materials and Methods
After receiving approval by the Animal Research Ethical Committee of Uppsala University, we included 18 pigs (mixed breed of Hampshire, Yorkshire, and Swedish landrace; mean weight, 29.5 +/- 7.6 kg) in the study. Six were used as a control group and were anesthetized for the same duration but without any form of surgery (control group). In six other pigs, a sternotomy was performed and heparin/protamine was given (sternotomy group). The six remaining pigs were treated as the sternotomy group but were also subjected to CPB (bypass group).
Premedication was given before transport with a neuroleptic (40 mg azaperone [Stresnil; Janssen, Beerse Belgium] given intramuscularly). General anesthesia was induced with 0.04 mg/kg atropine and 2.2 mg/kg tiletamine/zolazepam (Zoletil; Bayer AG, Leverkusen, Germany) given intramuscularly. A cannula was inserted in an ear vein and 5 micro gram/kg fentanyl was injected. A tracheotomy was performed and a cuffed endotracheal tube (inner diameter, 6 mm) inserted. Muscle relaxation was provided by 0.2 mg/kg pancuronium, and artificial ventilation was instituted using a Servo ventilator in the volume-cycled mode (Servo 900C; Siemens-Elema AB, Solna, Sweden). The tidal volume was 10 ml/kg at 20 breath/min with a positive end-expiratory pressure of 4 cmH2O, an inspiratory time of 25%, and an inspiratory pause of 10% of the inspiratory cycle. The inspired oxygen fraction (FIO2) was 0.5. Anesthesia was maintained with a constant infusion of a hypnotic (clomethiazole, 400 mg/h) combined with pancuronium (2 mg/h) and fentanyl (150 micro gram/h).
An 18-gauge catheter was inserted in the carotid artery via a cut-down for pressure measurements and blood sampling. A fiberoptic catheter (Pulsiocath 4F FT PV 2024; Pulsion Medical Systems, Munich, Germany) was inserted in the same artery and advanced into the aorta for lung water measurements. A Swan-Ganz thermodilution catheter was introduced in the external jugular vein via the same incision.
Sternotomy and Cardiopulmonary Bypass
In the pigs of the bypass and sternotomy groups, a median sternotomy was performed, and the pericardium and both pleura were opened. Heparin sodium (porcine type; 400 IU/kg) was administered. The activated clotting time was maintained at more than 400 s (Hemochron 400; International Technidyne Corp., Edison, NJ). In the bypass group, a single 28-French venous return cannula was inserted through the right atrial appendage and a 16-French cannula was inserted into the ascending aorta. The extracorporeal circuit consisted of a membrane oxygenator (Univox Membrane Oxygenation Module; Bentley; Baxter, Irvine, CA), a cardiotomy reservoir with filter (BCR-2500, Bentley; Baxter), and polyvinyl chloride tubing. Ringer's acetate solution (600 ml) and mannitol 15%(200 ml) were used to prime the circuit. Perfusion was conducted using a nonpulsatile pump (PMO 10–220 type; Gambro, Lund, Sweden). After the beginning of cardiopulmonary bypass, ventilation was stopped, the respirator disconnected, and the airway opened to the atmosphere. The aorta was clamped and cardioplegic solution (St. Thomas type I) with procaine (0.27 mg/l) injected in the root of the aorta until cardiac arrest. A minimum of 15 ml/kg was always given, and it was repeated every time cardiac mechanical activity was restarted (total: 21.1 +/- 4.0 ml/kg). Hypothermia to 30 [degree sign] Celsius was induced with a thermal exchanger (Chiller Thermo Circulator; Churchill Instrument Ltd., Perivale, UK) coupled to the oxygenator.
When the mean arterial pressure during CPB (51.7 +/- 8.3 mmHg) decreased to less than 40 mmHg for more than 5 min, a bolus of 50 micro gram epinephrine was given (this was only necessary in two pigs). The pump flow rate (70.2 ml [center dot] kg sup -1 [center dot] min sup -1) was limited by the venous return. To limit the hemodilution, no further crystalloid or blood was given. This also reduced the risk of microaggregate formation and trapping in the lung capillaries.
One hundred milliliters of the cardioplegic solution was given 10 min before the aortic clamp was released (total duration of the cardiac ischemia = 45 min). Rewarming was then initiated with the thermal exchanger and continued for 40 min. When the pigs had reached 37–38 [degree sign] Celsius (before bypass, 38–39 [degree sign] Celsius), the pump flow rate was progressively decreased during a few minutes. The pigs were then separated from the bypass. The total duration of cardiopulmonary bypass was 90 min. At the end of the bypass, all the pump prime was returned to the animals through the aortic cannula. Fifteen minutes before termination of bypass, ventilation was reinstituted at half the tidal volume with an FiO2of 1.0. Normal tidal volumes were given just before termination of CPB. The heparin effect was reversed in the sternotomy and the bypass groups with protamine (1 mg for each 100 IU used).
The control group had the same anesthesia performed as in the other two groups, but without cardiopulmonary bypass or any other form of surgery, and heparin and protamine were not given. The total duration of anesthesia was the same in the three groups. In the control group, three of the pigs were ventilated throughout the procedure with an FIO sub 2 of 0.5, whereas the the FIO2increased to 1.0 in the other three pigs during the last hour of anesthesia, which was similar to that in the two other groups. This enabled an analysis of the influence of FIO2on atelectasis and shunt formation.
The pigs of the three groups had a basal infusion of 150 ml/hour of NaCl 0.9%. In the two surgical groups (sternotomy and bypass groups), evaporation from the surgical field was estimated to 10–15 ml [center dot] kg sup -1 [center dot] h sup -1. Two hours with an opened sternum and the great vessels, the heart and the lungs in contact with the atmosphere may thus have caused a fluid loss of approximately 900 ml. In the bypass group, this fluid loss was more or less compensated for by the cardioplegic solution and the priming volume, which was returned to the pig at the end of the bypass. For the sternotomy group, this fluid loss was compensated for by infusion of 200 ml mannitol and Ringer's acetate during the period of the opened sternum.
The volume load was followed by repeated measurements of the hematocrit concentration.
Measurements consisted of arterial and mixed venous blood gases (ABL 300 and OSM 3 Hemoximeter; Radiometer, Copenhagen, Denmark), heart rate, systemic and pulmonary arterial pressures (series 7010 monitor; Tram, Marquette Electronics, Milwaukee, WI), cardiac output measured by thermodilution, extravascular lung water, and intrathoracic blood volume (measured using the double-indicator dilution method), respiratory mechanics, and ventilation-perfusion relations.
Measurements of Thoracic Intra- and Extravascular Fluid Volumes. The same indicator bolus was used to determine cardiac output and lung fluid volumes and consisted of indocyanine green (an intravascular marker) mixed in 5 ml ice-cold 5% glucose (a thermal intra- and extravascular indicator). The bolus was injected in the right atrium. The dilution curves for dye and temperature were recorded simultaneously in the aorta with the thermistor-tipped fiberoptic catheter. A lung water computer (Pulsion COLD Z-021; Pulsion Medical Systems) determined the mean transit time for the thermal indicator and for the dye indicator and calculated cardiac output, total thermal volume, intrathoracic blood volume, and extravascular lung water. All measurements were made in triplicate, and the mean was calculated and used for statistical evaluation (for further details, see Hachenberg et al. ).
Ventilatory Parameters: Compliance and resistance of the total respiratory system were measured using the technique of rapid airway occlusion during constant-flow inflation.  Resistance was calculated as the difference between peak airway pressure and the pressure at the end of a 2-s end-inspiratory pause, divided by the flow.  Compliance was calculated as tidal volume divided by the end-inspiratory pressure minus the end-expiratory pressure. Pressure and flow were measured in the ventilator on the inspiratory side and fed into a computer for on-line signal processing (Lab-VIEW 3.1 software; C-O Sjoberg Engineering, National Instruments, Austin, TX). Gas compression in the ventilator tubing was corrected for in the calculation of gas volume and flow. The mean value of two “inspiratory hold” maneuvers was used for statistical analysis.
Measurements of Ventilation-Perfusion Distribution: The technique (multiple inert gas elimination technique) is based on the steady-state elimination (obtained after a 40-min equilibration period) of six inert gases with different solubility in the blood (sulfur hexafluoride, ethane, cyclopropane, enflurane, diethylether, and acetone) as described by Wagner et al. [27–29] The mixture is dissolved in isotonic saline and infused at a constant rate (2–3 ml/min, depending on the minute ventilation) in a peripheral vein. After 40 min of infusion, under steady-state conditions, arterial and mixed venous blood were collected together with an expired gas sample and analyzed by gas chromatography (model 5890, series II; Hewlett-Packard, Little Falls, DE). These data allowed us to construct a virtually continuous distribution of blood flow or ventilation against ventilation-perfusion ratios (V with dot A/Q with dot), with separation of shunt (V with dot A/Q with dot < 0.005) from regions with low V with dot A/Q with dot ratios (0.0005 < V with dot A/Q with dot 0.1 = poorly ventilated lung units in relation to their perfusion) as well as normal V with dot A/Q with dot regions, units with high V with dot A/Q with dot ratios (10 V with dot A/Q with dot < 100) and dead space (V with dot A/Q with dot > 100). The dispersion of V with dot A/Q with dot ratios is expressed as the logarithmic standard deviation of perfusion distribution (log SDQ). It describes the degree of V with dot A/Q with dot mismatch.
Computed Tomographic Scans. At the end of the study, the pigs were moved to the CT scan laboratory (Somaton Plus 4; Siemens, Munchen, Germany). A frontal scoutview covering the chest was obtained at end-expiration to define the limits of the lungs. One CT scan was performed at end-expiration, 0–1 cm above the diaphragm (146 mA, 140 kV, 5-mm slice thickness). The CT scans were analyzed for distribution of lung tissue density: the total lung area was delineated manually. Nonaerated lung tissue (atelectasis) was defined as regions with attenuation values between -100 and +100 Hounsfield units  and poorly aerated lung tissue as regions with values between -500 and -100 Hounsfield units. The extent of atelectasis and poorly aerated lung tissue was expressed as a percentage of the total lung area (excluding the mediastinum).
Baseline Measurements. A delay of 30 min was allowed after the surgical preparation before baseline measurements were made.
Postbypass or Second Measurements (3 h after Baseline). The infusion of inert gases was restarted after the end of the bypass and due to the time necessary to obtain a new steady state, the postbypass measurements of circulatory and ventilatory variables were not taken until 45 min after separation from bypass. During this waiting time, the CPB cannulas were removed and the chest closed.
To detect any specific effect of the injection of protamine, two other series of measurements were made, one just before and the other 10 min after the end of the injection. It consisted of hemodynamics, blood gases with shunt calculation (standard formulae), and lung mechanics.
At the end of the experiment, the animals were killed with an intravenous injection of potassium chloride, and the lungs were removed for inspection.
Data in the text, tables, and figures are presented as means +/- SD. Baseline comparisons were made with an unpaired Student's t test as for nonrepeated values, which were only measured once after bypass; that is, for atelectasis and poorly aerated lung tissue. The effect of bypass or sternotomy alone compared with the control group was calculated by analysis of variance for repeated measures. Simple regression analysis was performed to test the correlation between two parameters of all groups pooled together. P < 0.05 was considered significant.
The pigs were of the same size in the three groups (control group, 31.2 +/- 10.7 kg; sternotomy group, 31 +/- 2.8 kg; and bypass group, 27.8 +/- 2.6 kg). There were no significant differences in ventilation, respiratory mechanics, central hemodynamics, or gas exchange (Table 1, Table 2, Table 3).
Effect of Time, Sternotomy, and Cardiopulmonary Bypass
Ventilation and Respiratory Mechanics. The ventilator settings were kept essentially constant during the experiment so that minute ventilation was not significantly different compared with baseline or between the groups (Table 1). After the bypass period, peak and end-inspiratory airway pressures were significantly increased and compliance decreased, whereas no changes were seen in the two other groups (Table 1).
Hemodynamics and Lung Fluids. No significant changes were noted in the cardiac output (Table 2). Mean pulmonary artery and central venous pressures increased after the cardiopulmonary bypass but not in the two other groups. No changes in lung water or intrathoracic blood volume were seen in any group (Table 2).
Ventilation-Perfusion Relations and Gas Exchange. Intrapulmonary shunt increased markedly after bypass (to 18%; P < 0.01), but less so in the sternotomy group (to 10%; P < 0.01) and not at all in the controls (3.5%)(Table 3;Figure 1). The difference between the sternotomy group and the control group was not significant, whereas it was significant between the bypass group and the two other groups. There was only minor perfusion of “low V with dot A/Q with dot” regions and no increase after bypass. In all groups, log SDQ was unaffected compared with baseline, and there was no difference between the groups (bypass group, 0.70 +/- 0.21; sternotomy group, 0.71 +/- 0.11; controls, 0.67 +/- 0.08). Partial pressure of oxygen (PaO2) was significantly reduced after the bypass but not so in the sternotomy group (Table 3). There was a strong inverse correlation between shunt, measured by the multiple inert gas elimination, and PaO2(R2= 0.75; P <0.0001; n = 18: data from all three groups). Adding “low V with dot A/Q with dot” to the regression analysis did not improve the correlation. Dead space, measured by the multiple inert gas elimination, was not altered after the bypass. However, PaO2was significantly increased after the bypass, despite maintained minute ventilation and dead space. No change in PaO2was seen in the two other groups (Table 3). There was a good correlation between PaO2and shunt (R2= 0.64).
Atelectasis and Lung Aeration. There was only a small amount of atelectasis (approximately 1%) in the control group, despite more than 4 h of anesthesia in the supine position (dorsal recumbency)(Table 3, Figure 2(A)). The pigs in the sternotomy group showed more atelectasis (4%)(Figure 2(B)) than the controls, and the pigs in the bypass group showed extensive atelectasis, accounting for 35% of the total lung area (P < 0.05 compared with the other groups)(Figure 2(C)). Poorly aerated lung tissue amounted to approximately 20–30%, with no significant difference between the groups (Table 3). There was a good correlation between shunt and atelectasis (R2= 0.67; P <0.0001;Figure 3), and also a negative correlation between PaOsub 2 and atelectasis (R2= 0.53; P < 0.001) and between PaOsub 2 and atelectasis (R2= 0.63).
In the control group, no differences were seen in the gas exchange parameters or in the amount of atelectasis among the three pigs ventilated with 100% oxygen compared to the three pigs ventilated with 50%. Because of the small number of pigs in each subgroup, we did not perform any statistical analyses.
No differences were noted in any of the parameters measured 10 min after the end of the injection of protamine when compared with the measurements made just before the injection (mean pulmonary arterial pressure before, 15 +/- 1.4 mmHg vs. 16 +/- 1.9 mmHg after protamine injection; shunt, 13.1 +/- 3.3% vs. 10.8 +/- 3.5%; PaO2/FIO sub 2, 448 +/- 55 mmHg vs. 477.3 +/- 59.5 mmHg; and peak airway pressure, 16.1 +/- 3.1 cmH2O vs. 16.5 +/- 2.9 cmH2O).
In this pig model study, we found a large increase in intrapulmonary shunt and a decrease in PaO2after cardiopulmonary bypass, which is reasonably explained by the formation of large atelectasis (Figure 2(C)). Such changes were not seen in a control group that was studied for the same time of anesthesia but without bypass or any other surgery (Figure 2(A)). This supports our hypothesis that atelectasis persisting from the bypass period, or created after operation, is a major cause of the intrapulmonary shunt and decrease in oxygenation commonly seen in persons having cardiac surgery. Intrapulmonary shunt was also moderately increased in a sternotomy group exposed to the same surgery as the bypass group but without the CPB. More atelectasis developed in these pigs (Figure 2(B)) compared with the controls. This indicates that not only the bypass but also the surgery (although to lesser degree) are responsible for the formation of atelectasis.
Cardiopulmonary Bypass and Atelectasis
Even by 1958 Dodrill  had described alveolar collapse apparent on microscopic examination in some cases of post-CPB fatal respiratory failure. However, varying results have been obtained by conventional chest roentgenogram. Atelectasis was seen in 20% of the patients after CPB in a study by Rolla et al.,  and up to 64% in another study by Gale et al.  No correlation between atelectasis and shunt or degree of hypoxemia could be demonstrated in any of these studies. Hachenberg et al.  used CT scanning in patients in respiratory failure after CPB. They found in this selected group of patients that atelectasis amounted to 24% of the lung area near the diaphragm where the exposure was made, and that shunt, mean 26%, correlated with the degree of atelectasis. In our study, large atelectasis appeared in all bypass animals. Furthermore, shunt and PaO2correlated well with the atelectasis. The shunt could be explained to as much as 67% by the atelectasis, which thus was the major cause of the impaired oxygenation (Figure 3). The improved detection and quantification of atelectasis by computed tomography may explain why correlations between atelectasis and shunt or oxygenation were seen with CT but not earlier with conventional roentgenograms. It is tempting to extrapolate the findings to human conditions after CPB.
Sternotomy and Shunt
The shunt was also increased in the sternotomy group but was significantly lower than in the bypass group and was not significantly different from that seen in the controls.
The formation of atelectasis in the sternotomy group may be an effect of the bilateral pleurotomy, pleural effusion, and even lung trauma during surgery and insertion of the chest tubes. Indeed, pleurotomy is sometimes associated with a decrease in lung function and gas exchange [32,33] but sometimes not. [31,34]
Despite the increase in shunt, no clear decrease in arterial oxygenation was seen in the sternotomy group. This may be explained by an increase in mixed venous oxygen pressure that attenuates the influence of shunt on arterial oxygenation. Indeed, calculations from the inert gas data showed that oxygen uptake was significantly reduced (from 146 +/- 29 to 122 +/- 23 l/min; P < 0.01). This decrease in oxygen consumption, which was not seen in the other two groups, may be explained by thermal loss through the open chest. The active rewarming of the pigs in the bypass group at the end of the CPB may have prevented the same decrease in metabolism that occurred in this group.
Other Mechanisms of Gas Exchange Impairment
Other causes of hypoxemia after sternotomy and CPB have also been proposed. In 1977, Pennock  reviewed factors that may cause postbypass respiratory failure: microembolism, interstitial edema, hypoxia of the parenchyma of the lung, and oxygen toxicity. Bubble oxygenators have been incriminated, but no or little advantage in using membrane oxygenators have been shown. [8,35] Lindberg et al.  did not find any difference between pulsatile and nonpulsatile flow on pulmonary dysfunction parameters. Others have postulated that leukocyte activation is a major cause of the postbypass respiratory insufficiency. [9,10,36,37] Gillinov et al.  showed that pretreatment with a potent inhibitor of complement activation, in a pig model, decreased pulmonary hypertension. However, the complement inhibition did not improve post-CPB oxygenation and had no significant effect on leukocyte kinetics or lung histologic features. A decrease in the surfactant activity has also been incriminated,  but more than 4 h of cardiopulmonary bypass was necessary to detect a decrease in the action of the surfactant in dogs.  An increase in the extravascular lung water was found,  but its early return to pre-bypass values suggests that mechanisms other than changes in lung water influence gas exchange.  Others have shown that CPB is associated with a significant increase in the permeability of the alveolar-capillary barrier, but without correlation to the increase in P sub (A-a)O2 unless CPB has been markedly prolonged.  Finally, MacNaughton et al.  did not find any increase in pulmonary endothelial permeability, and they hypothesized that the major component of the deterioration in lung function was probably atelectasis occurring during bypass.
Fluid Balance and Lung Tissue Edema
A positive fluid balance may enhance the shunt by vascular dilatation and edema formation and has been considered a mechanism of shunt in CPB.  However, in the present study, fluid administration was limited and the extravascular lung water was not significantly increased. The net fluid balance seems to have been more or less equal in all groups, as shown by the hematocrit values, which did not differ between the groups. This allows us to conclude that the densities seen on the CT scans were not interstitial edema but rather lung collapse caused by a mechanism other than fluid. In addition, to counter hemodilution, we decided to limit as much as possible the priming volume and not to add Ringer's solution during CPB. This may explain why the blood flow during CPB was low and why the pigs were acidotic at the end of CPB. Acidosis may have some deleterious effect on post-CPB myocardial function, but we found no difference in the cardiac output among the groups.
In conclusion, cardiopulmonary bypass caused large atelectasis in a porcine model (Figure 2(C)). The atelectasis seems to explain most of the large shunt and impaired oxygenation of blood that was seen (Figure 3). However, thoracic surgery may also be responsible, but to a lesser degree, for lung collapse during cardiac surgery as shown by the presence of small areas of atelectasis in the sternotomy group. Pigs receiving mechanical ventilation for a similar period without cardiopulmonary bypass developed no or only minor atelectasis (Figure 2(A)) and no or only minor gas exchange impairment. The similarities among the groups in the anesthetic procedure, mechanical ventilation, and surgical procedure (for the bypass and sternotomy groups) point to the period of no ventilation and absence of perfusion during the bypass as the major insult causing lung collapse. Thus atelectasis may be one important cause of postbypass gas exchange impairment.