Within-breath arterial PO2 oscillations in an experimental model of acute respiratory distress syndrome

E. M. Williams1, J. P. Viale2, R. M. Hamilton1, H. McPeak1, L. Sutton1 and C. E. W. Hahn1

1Nuffield Department of Anaesthetics, University of Oxford, Radcliffe Infirmary, Oxford OX2 6HE, UK. 2Service Anesthésie Réanimation, Hopital de la Croix Rousse, 103 Grande-Rue de la Croix, Rousse, F-69317 Lyon Cedex 04, France*Corresponding author

Accepted for publication: April 10, 2000


    Abstract
 Top
 Abstract
 Introduction
 Methods and results
 Comments
 References
 
Tidal ventilation causes within-breath oscillations in alveolar oxygen concentration, with an amplitude which depends on the prevailing ventilator settings. These alveolar oxygen oscillations are transmitted to arterial oxygen tension, PaO2, but with an amplitude which now depends upon the magnitude of venous admixture or true shunt, Q·S/Q·T. We investigated the effect of positive end-expiratory pressure (PEEP) on the amplitude of the PaO2 oscillations, using an atelectasis model of shunt. Blood PaO2 was measured on-line with an intravascular PaO2 sensor, which had a 2–4 s response time (10–90%). The magnitude of the time-varying PaO2 oscillation was titrated against applied PEEP while tidal volume, respiratory rate and inspired oxygen concentration were kept constant. The amplitude of the PaO2 oscillation, {Delta}PaO2, and the mean PaO2 value varied with the level of PEEP applied. At zero PEEP, both the amplitude and the mean were at their lowest values. As PEEP was increased to 1.5 kPa, both {Delta}PaO2 and the mean PaO2 increased to a maximum. Thereafter, the mean PaO2 increased but {Delta}PaO2 decreased. Clear oscillations of PaO2 were seen even at the lowest mean PaO2, 9.5 kPa. Conventional respiratory models of venous admixture predict that these PaO2 oscillations will be reduced by the steep part of the oxyhaemoglobin dissociation curve if a constant pulmonary shunt exists throughout the whole respiratory cycle. The facts that the PaO2 oscillations occurred at all mean PaO2 values and that their amplitude increased with increasing PEEP suggest that Q·S/Q·T, in the atelectasis model, varies between end-expiration and end-inspiration, having a much lower value during inspiration than during expiration.

Br J Anaesth 2000; 85: 456–9

Keywords: ventilation, positive end-expiratory pressure; oxygen, tension


    Introduction
 Top
 Abstract
 Introduction
 Methods and results
 Comments
 References
 
Krogh and Lindhard, in 1914,1 hinted that alveolar and arterial oxygen tensions, PAO2 and PaO2 respectively, could fluctuate during the respiratory cycle. Bergman, in 1961,2 confirmed this hypothesis with an animal model using arterial saturation, SaO2, as a proxy for PaO2. Although Bergman was not able to measure PaO2 continuously, he hypothesized that fluctuations in SaO2 during the respiratory cycle might be due to the pulmonary shunt fraction (Q·S/Q·T) changing between the inspiratory and expiratory phases of the respiratory cycle. Further isolated studies over the past 30 yr have shown that oscillations in arterial oxygen tension, cotemporaneous with the respiratory cycle, do occur and that they are most apparent in the presence of hypoxaemia or significant venous admixture.3 4 The origin of these oscillations, with peak-to-peak amplitude denoted by {Delta}PaO2, are thought to be in the complex interaction between the continuous process of gas exchange and the tidal process of the ventilation respiratory cycle.57 We considered that if variations in Q·S/Q·T caused by atelectasis occurring during expiration and alveolar reopening during inspiration, did occur in the lung, then the amplitude of the oscillations would be sensitive to the application of PEEP when both the inspired oxygen fraction (FIO2) and the ventilator settings were kept constant. We assessed the effect of PEEP on {Delta}PaO2 in an experimental model of atelectasis [simulating acute respiratory distress syndrome (ARDS)] which developed moderate to severe levels of pulmonary shunt.


    Methods and results
 Top
 Abstract
 Introduction
 Methods and results
 Comments
 References
 
In seven adult female dogs (weight range 10.4–13.3 kg), anaesthesia was induced by i.v. pentobarbitone sodium (60 mg ml–1) and maintained by infusion (60–180 mg h–1). The surgical preparation was performed as described elsewhere.8

A prototype intravascular PO2 sensor (IE Sensors, Salt Lake City, Utah, USA) was inserted into the aorta and left to stabilize. This Clark-type amperometric sensor had a 10–90% response time of 2–4 s. Fibre optic pulmonary artery catheters (Opticath Catheter; Abbott Critical Care Systems, Chicago, Illinois, USA) were used to measure mixed venous blood oxygen saturation continuously using the principle of reflectance spectroscopy (Oximetric 3; Abbott Critical Care Systems). Intermittent measurement of blood gases and pH were made using a blood-gas analyser (ABL 330; Radiometer, Copenhagen, Denmark), and the same blood sample was used to measure other blood variables by co-oximetry (OSM 3, Radiometer). Cardiac output, Q·T, was measured by thermodilution (Oximetric 3, Abbott Critical Care Systems). The value of Q·S/Q·T was calculated conventionally using the FIO2, blood gas and saturation data, assuming a respiratory quotient of 0.8.

Correct positioning of the PO2 sensor in the aorta was assessed by observing the blood pressure tracing from the sensor’s blood sampling port and by a stable mean PaO2. The PO2 sensor was calibrated against drawn heparinized arterial blood using the blood-gas analyser. Inspired and expired oxygen concentrations were measured, at the end of the tracheal tube, using a respiratory Quadrupole mass spectrometer (VG Quadrupoles, Middlewich, UK).

Throughout the study, the mechanically ventilated animals received a constant mean (SD) inspired FIO2 of 0.72 (0.01). Tidal volume and respiratory rate were kept constant during each individual study, but ranged from 0.25 to 0.30 litres and 11 to 13 breaths min–1, respectively, between animals to maintain an end-expired carbon dioxide concentration of 4.7 (0.6)% v/v. After surgery, a stabilization period of 1–2 h was allowed and baseline (prelavage) values were recorded (Table 1). Bronchopulmonary lavage was performed to deplete the lung surfactant, inducing a large pulmonary shunt. The lavage procedure was repeated, typically 10–20 times, until PaO2 had decreased from 57 to 10 kPa (Table 1). This procedure produced a typical initial shunt fraction, Q·S/Q·T, of 53% (SD 16%).


View this table:
[in this window]
[in a new window]
 
Table 1 The effect of increasing (upper of each pair of rows) and decreasing (lower row) PEEP on blood gases, oxygen saturation, peak airway pressure, shunt fraction, cardiac output and arterial blood pressure. The mean (SD) is shown for seven animals unless indicated by a superscript number. Blood-gas values were corrected to 38°C. Statistical significance was assessed using the paired Student’s t-test (unpaired when n not equal). *Prelavage value different (P<0.05) from the postlavage value; **value of the variable at 0 PEEP significantly (P<0.05) altered by 2 kPa PEEP; ***Value of the variable during lung inflation at a given PEEP level significantly different (P<0.05) from the value during deflation. Otherwise no statistical differences were found
 
When the mean PaO2 was stable, a set of observations was taken, followed by simultaneous venous and arterial blood samples. The time-varying PaO2 signal was recorded over 30 breaths by computer. This procedure was repeated every 20 min with 0.5 kPa increments of PEEP from 0 to 2 kPa, followed by a return to zero PEEP (ZEEP) in decrements of 0.5 kPa.

Results (Table 1 and Fig. 1) are expressed as mean (SD), and the statistical analysis was by one-way analysis of variance with repeated measurements. When this analysis showed significance, post hoc comparisons of the means were made by Scheffé’s test. A P value less than 0.05 was considered as significant. Table 1 shows the overall effect of increasing and decreasing PEEP on blood gases, oxygen saturation, cardiac output, shunt fraction and arterial blood pressure. The table also shows the conventional pulmonary blood-flow shunt fraction calculated at each PEEP level. Bronchopulmonary lavage reduced the mean (prelavage) PaO2 by 47 kPa and the mean PvO2 by 2.3 kPa. At the initial low mean PaO2, 9.5 (0.7) kPa, the PaO2 signal began to oscillate about its mean value with an amplitude of 1.2 (0.8) kPa, and both the mean PaO2 and its oscillation amplitude began to increase as PEEP was imposed. Figure 1A shows typical results taken from a single animal study. It must be noted that the PaO2(t) time-varying tracings presented in this figure are not related to each other on the time axis, and no physiological inference can be made from the time-phase differences of these traces. Figure 1B shows the effect of incremental changes on {Delta}PaO2 (mean (SD)) for all the studies, as PEEP was firstly increased from 0 to 2 kPa and then decreased from 2 to 0 kPa. Each individual study showed the same effects of PEEP on {Delta}PaO2 as those illustrated in Fig. 1, namely that (i) as PEEP was initially increased, both the mean PaO2 and {Delta}PaO2 increased; but that (ii) {Delta}PaO2 began to decrease after a certain PEEP value was reached, although the mean PaO2 continued to rise. This pattern was reversed as PEEP was reduced back to its baseline (ZEEP) level. In all instances, the PaO2 oscillations followed the ventilator frequency, and appeared to become maximal at the end of inspiration and minimal at the end of expiration. There was also a transport lag between the ventilator inspiratory–expiratory phases and the PaO2 oscillations.



View larger version (25K):
[in this window]
[in a new window]
 
Fig 1 (A) Breath-by-breath oscillations in PaO2 (with peak-to-trough amplitude {Delta}PaO2) at different levels of PEEP, and at a constant FIO2 of 0.72, recorded by an intravascular PaO2 sensor with a 10–90% response time of 2–4 s. The PaO2 is at its peak at the end of inspiration and at its lowest at the end of expiration. Data from a single animal are shown. The respiratory rate was 11 breaths min–1, and the tidal volume was kept constant at 310 ml. There is no phase (time) relationship between any of the above time-varying PaO2 traces shown at the various PEEP levels. (B) The effects of incremental changes in PEEP on {Delta}PaO2 from 0 to 2 kPa and decremental PEEP from 2 to 0 kPa. The mean (SD) for seven animals is shown.

 

    Comments
 Top
 Abstract
 Introduction
 Methods and results
 Comments
 References
 
There were two main findings of this study. First, an experimental animal model of ARDS was associated with significant oscillations in PaO2 linked with the respiratory cycle. Secondly, the peak-to-peak amplitude of these fluctuations, {Delta}PaO2, was dependent upon the level of PEEP applied to the damaged lungs, the magnitude of the amplitude increasing up to 3 kPa with increasing PEEP. The true in vivo PaO2 oscillations were probably considerably greater than this because the intravascular PaO2 sensors had a 10–90% response time of only 2–4 s. This would have attenuated the rapidly changing true in vivo arterial PO2 signals. A rough calculation of the true magnitude of the PaO2 oscillations can be derived from the knowledge of the PO2 sensor response time, the respiratory rate, by approximating the inspiration:expiration ratio to 1:1 and then using the formula of Arieli and Van Liew9 to calculate the attenuation of the physiological PaO2 signal due to the sensor membrane. This rough calculation suggests that the in vivo PaO2 peak-to-trough oscillations are up to three times greater than those recorded by the IE Sensors Inc. intravascular PO2 sensors, shown in Fig. 1A.

Both Purves and the Kreuzer group, in the 1960s and 1970s, reported respiratory-induced oscillations in arterial PO2 in animal models.3 4 These were a direct consequence of tidal ventilation fluctuations in alveolar gas oxygen tension, and had the same period as the ventilator setting. An increase in tidal volume or a decrease in ventilation frequency led to an increase in the amplitude of the PaO2 oscillations. These early studies ruled out the possibility that the effects of cyclical variations in arterial pressure, or blood flow, could produce the measured PaO2 fluctuations. In contrast to our own animal ARDS model, these earlier studies investigated these PaO2 oscillations only in the healthy lung.

Our own findings confirm those of previous authors, such as Bergman’s report in 19612 of oscillations in arterial oxyhaemoglobin saturation, SaO2, (when haemoglobin was not fully saturated) and, more recently, those of Elwell et al. in 1996,11 which demonstrated that oscillations in arterial saturation occur with the induction of mild hypoxia. The results of Elwell et al. were explained by Lovell et al. in 1997,11 who showed that changes in ventilator settings could alter the SaO2 oscillations. Bergman showed that the amplitude of the SaO2 oscillations diminished as haemoglobin became saturated, but he was not able to measure PaO2 on-line.2 However, on the basis of his SaO2 studies, he hypothesized that the most likely explanation of his results was that pulmonary shunt varied during the respiratory cycle as a result of the lung collapsing during expiration and then reopening during positive-pressure inspiration.

Two problems need to be faced. These are that (i) the {Delta}PaO2 amplitudes observed in our study were larger than would be expected in healthy lungs; and (ii) current knowledge suggests that PaO2 oscillations on the steep part of the oxyhaemoglobin association/dissociation curve are buffered by the shape of the curve describing the relationship of oxygen content with PO2. Significant PaO2 oscillations in this steep part of the curve could be caused by: (i) the presence of inhomogeneity in the lung ventilation–perfusion ratio induced by the pulmonary lavage; (ii) different degrees of atelectasis occurring in the lung during the inspiratory and expiratory phases of the respiratory cycle (the Bergman hypothesis); or (iii) a combination of both mechanisms, as they do not exclude one another.

We consider that, during inspiration, more alveoli are recruited, with a decrease in venous admixture (during inspiration) and consequently an increase in both the mean PaO2 and the PaO2(t) time-varying oscillatory signal. During expiration, these recruited alveoli could collapse and contribute to an increased shunt fraction and, thus, a decrease in mean PaO2. When PEEP was increased sufficiently to induce permanent recruitment during the expiratory phase, the overall PaO2 amplitude would decrease and the mean PaO2 rise. On the other hand, if shunt fraction was constant during both the inspiratory and expiratory phases of respiration (the conventional view of Q·S/Q·T) then the {Delta}PaO2 oscillations would not appear at all at low mean PaO2 values because of the buffering capacity of oxyhaemoglobin in this region of the dissociation curve.

It now seems clear that PaO2 oscillations occur in the atelectatic lung, and that the application of PEEP not only elevates the mean arterial PaO2 but also affects the magnitude of the PaO2 oscillations superimposed on this mean. The effect of these oscillations in the clinical care setting is not clear.


    Acknowledgements
 
This study was supported by the Medical Research Council (MRC) and the Wellcome Trust. R. M. H. received an MRC Research Studentship. J. P. V. acknowledges the receipt of Grant DA 1896 MERS.


    References
 Top
 Abstract
 Introduction
 Methods and results
 Comments
 References
 
1 Krogh A, Lindhard J. On the average composition of the alveolar air and its variation during the respiratory cycle. J Physiol 1914; 47: 431–45

2 Bergman NA. Cyclic variations in blood oxygenation with the respiratory cycle. Anaesthesiology 1961; 22: 900–8[ISI]

3 Purves MJ. Fluctuations of arterial oxygen tension which have the same period as respiration. Respir Physiol 1966; 1: 281–96[ISI][Medline]

4 Folgering H, Smolders FDJ, Kreuzer F. Respiratory oscillations of the arterial PO2 and their effects on the ventilatory controlling systems in the cat. Pflügers Arch 1978; 375: 1–7

5 Hlastala MP. A model of fluctuating alveolar gas exchange during the respiratory cycle. Respir Physiol 1972; 15: 214–32[ISI][Medline]

6 Yokota H, Kreuzer F. Alveolar to arterial transmission of oxygen fluctuations due to respiration in anaesthetised dogs. Pflügers Arch 1973; 340: 291–306

7 Lin KH, Cumming G. A model of time-varying gas exchange in the human lung during a respiratory cycle at rest. Respir Physiol 1973; 17: 93–112.[ISI][Medline]

8 Williams EM, Hamilton R, Sutton L, Hahn CEW. Measurement of respiratory parameters using inspired oxygen sinusoidal forcing signals. J Appl Physiol 1996; 81: 998–1006[Abstract/Free Full Text]

9 Arieli R, Van Liew HD. Corrections for the response time and delay of mass spectrometers. J Appl Physiol 1981; 51, 1417–22[Abstract/Free Full Text]

10 Elwell CE, Owen-Reece H, Wyatt JS, Cope M, Reynolds EOR, Delpy DT. Influence of respiration and changes in expiratory pressure on cerebral haemoglobin concentration measured by near infrared spectroscopy. J Cereb Blood Flow Metab 1996; 16, 353–7[ISI][Medline]

11 Lovell AT, Owen-Reece H, Elwell CE, Smith M, Goldstone JC. Predicting oscillation in arterial saturation from cardiorespiratory variables. Implications for the measurement of cerebral blood flow with NIRS during anaesthesia. Adv Exp Med Biol 1997; 428: 629–38[ISI][Medline]