INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Impairment of intrapulmonary oxygen transport as indicated by an increased alveolar-arterial oxygen gradient ([A-a]DO₂) may be caused by an increased diffusion distance for oxygen in the lungs or by an increase in venous admixture. However, particularly in patients with severe impairment of lung function such as pneumonia or acute respiratory distress syndrome (ARDS), an "oxygen steal" (i.e., a measurable consumption of oxygen by lung tissue itself) has been reported (1). Previous studies comparing measurements of oxygen consumption (O₂) with the reverse Fick method (O_2Fick) or indirect calorimetry (O_2cal) found, that O_2cal was significantly greater than O_2Fick (2, 3). Although O_2cal is assumed to measure total body O₂, the reverse Fick method excludes intrapulmonary O₂ (O_2pulm) from the measurement (4). Therefore, it was concluded and has now been widely accepted that the difference in values provided by the two methods represents the amount of oxygen consumed by the lungs (5). However, because no "gold standard" exists for measuring O_2pulm under clinical conditions, and considering the inherent methodologic weaknesses particularly of the reverse Fick method, it is important to standardize both calorimetric and reverse Fick measurements of O₂ as far as possible.

Data from human subjects breathing under normal physiologic conditions are not currently available. In humans and animals with pneumonia, a considerable increase in O_2pulm has been reported (4, 6). In addition, the ability of infected lung tissues to extract oxygen has been shown in isolated lung preparations (7). However, the underlying mechanisms of this lung oxygen consumption are unclear. It is unknown whether intra- or extrapulmonary structures are mainly responsible for the oxygen steal that occurs during pneumonia. It has been suggested that the increased O₂ of infected lungs is caused by increased oxidative metabolism in neutrophils and macrophages lodged in the pulmonary microcirculation, associated with an increase in cell number and phagocytosis, with the formation of oxygen-derived free radicals and an increase in oxidative degradative processes such as lipid peroxidation (1, 5).

Another hypothesis for the measurable increase in O_2pulm during pneumonia is an increase in oxygen-dependent metabolic processes within injured lung tissues, such as oxidative phosphorylation using glucose, lactate, pyruvate, and amino acids as potential substrates, an increased synthesis of eicosanoids, or the inactivation of vasoactive amines by oxygen-consuming enzymes such as monooxidases (8).

Because physiologically significant consumption of oxygen by the lung is still a controversial issue, the first aim of the present study was to define both this existence of this phenomenon as well as its magnitude. A second aim was to investigate whether there is a correlation between the extent of pulmonary inflammation and the magnitude of O_2pulm in patients with pneumonia. Transpulmonary lactate and glucose flux was calculated to test whether the measured values of O_2pulm in such patients are associated with metabolic processes accompanied either by increased lung lactate or glucose utilization or by increased lung lactate or glucose production, respectively. The cellularity of bronchoalveolar lavage fluid (BALF) was determined to show how activated neutrophils and macrophages may contribute to increased O_2pulm.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Selection

The study was approved by the local ethics committee, and informed consent was obtained from the subjects' next of kin. Seventy-five critically ill patients (31 female, 44 male) admitted to the intensive care unit (ICU) of the University Hospital Charité were included.

Inclusion criteria were a requirement for mechanical ventilation for at least 3 d, the need for right heart catheterization, and the need for arterial and mixed venous gas measurements for routine clinical management. Endotracheal intubation and mechanical ventilation were performed in cases of loss of consciousness, hemodynamic instability, inadequate muscle strength, and/or inadequate ventilation (positive end-expiratory pressure [PEEP]  5 cm H₂O, breathing rate > 30 breaths/min), as well as in cases of inadequate gas exchange (Pa_O2  60 mm Hg/FI_O2 = 0.6; Pa_CO2 > 55 mm Hg).

The patients were divided for analysis into two groups according to the existence (n = 41) or absence (n = 34) of pneumonia.

According to the International Consensus Conference on the clinical investigation of ventilator-associated pneumonia (VAP) (9), this condition was diagnosed when new, progressive, or persistent infiltrates; purulent tracheal secretions; and one of the following signs were found: quantitative culture of lower respiratory tract secretions obtained by a technique that minimizes contamination with upper respiratory tract flora (bronchoalveolar lavage [BAL]), positive blood culture of an organism found in lower respiratory tract secretions, positive pleural fluid culture of an organism identical to the organism found in lower respiratory tract secretions, histologic proof of VAP on open-lung biopsy or at autopsy (abscess formation, or consolidation with polymorphonuclear leukocyte (PMN accumulation), plus culture of > 10⁴ microorganisms/g of lung tissue. The extent of pulmonary infiltration was quantified by means of radiographic classification (number of infected lobes) using chest X-ray or computed tomography (CT), respectively. All patients who did not fulfill these criteria but nevertheless had clinical signs of an acute inflammatory intrapulmonary process were excluded from the study. Therefore, absence of pneumonia was defined as the absence of all of the criteria mentioned.

Healed pneumonia was defined if no radiographic (chest X-ray, CT scan) evidence of pulmonary infiltrate, purulent tracheal secretions, or quantitative culture of lower respiratory tract secretions was found on at least three consecutive days. These patients in the pneumonia group were then studied a second time.

Clinical Status and Lung Function

The clinical status of the patients was defined by the Acute Physiology and Chronic Health Evaluation (APACHE) II scoring system (10).

The extent of acute lung injury was assessed during each set of measurements from the Murray score (11), and the extent of venous admixture (shunt fraction S/T) and (A-a)DO₂ were calculated. The Murray score is based on the results of chest radiography, the calculation of gas exchange parameters (oxygenation index: Pa_O2/FI_O2), the use of PEEP, and the degree of respiratory compliance. S/T and (A-a)DO₂ were calculated with standard equations (1, 8).

In patients with accumulation of extravascular lung water (EVLW) as shown by chest radiography (n = 16), a fiberoptic thermistor catheter was inserted via a radial or femoral artery. Subsequently, EVLW was measured by a technique involving simultaneous dye and thermal indicator dilution measurements (COLD Z-021; Pulsion, München, Germany) (12).

O₂ Measurements

Controlled ventilation was used for each patient, with the same mechanical ventilator (Model 7200ae; Puritan-Bennett Corporation, Carlsbad, CA). The frequency and minute volume were adjusted to maintain Pa_CO2  40 mm Hg. During the measurements, patients were ventilated with an FI_O2  0.8. No changes in minute volume, PEEP, or FI_O2, and no manipulations such as tracheal suctioning or nursing interventions, were allowed within 30 min before measurements were made. The procedure was discontinued if there was a decrease in Sa_O2 monitored by means of pulse oximetry (Solar 8000; Marquette-Hellige Medical Systems, Milwaukee, WI) to < 90%. To achieve stable conditions during measurements, patients were paralyzed with intermittent bolus injections of vecuronium bromide (0.1 mg/kg). Analgosedation was maintained by continuous intravenous infusion of midazolam (0.1 mg/kg/h) and fentanyl (50 µg/kg/h).

O_2cal was measured with the Puritan-Bennett 7250 metabolic monitor operating in conjunction with the mechanical ventilator. This indirect calorimeter has been described and validated in detail elsewhere (13).

Oxygen consumption according to the reverse Fick method was calculated both with continuous cardiac output measurement (OptiQVUE Monitoring system; Abbott, North Chicago, IL) and conventional cardiac output (CO) measurement by thermodilution, with bolus injections of 10 ml normal saline at room temperature (Oximetrix Monitor; Abbott). O_2Fick (ml/min/m²) was calculated from the following equations:

(1)

(2)

(3)

where Ca_O2 (Cv_O2) (ml O₂/100 ml) is arterial (mixed venous) oxygen content, Pa_O2 (Pv_O2) (mm Hg) is arterial (mixed venous) oxygen tension, Hb (g/100 ml) is the hemoglobin content of blood, and Sa_O2 (Sv_O2) (%) is the arterial (mixed venous) oxyhemoglobin saturation. In accord with these measurements, right heart catheterization was performed using an OPTI-Q CCO-catheter (Abbott) or Opticath-catheters (P7110-E; Abbott). The distal lumen was used for mixed venous blood sampling. Radial artery catheterization (Vygon catheter, Ecoven, France) was used for arterial blood sampling. Blood hemoglobin saturation and hemoglobin concentration were measured with a cooximeter (OSM-3; Radiometer, Copenhagen, Denmark). Blood gas analyses were performed in a blood gas analyzer (ABL-505; Radiometer). Blood concentrations of glucose and lactate were measured with a glucose-lactate-analyzer (YSI 2300 Stat Plus; Yellow Springs Instrument, Yellow Springs, OH).

All patients were studied on three consecutive days. On each day, from two to four serial determinations (sets of measurements) of both O_2Fick and O_2cal were made simultaneously. Serial measurements were started when body core temperature and hemodynamic variables were considered stable. All values for O₂ were corrected to standard temperature and pressure, dry (STPD). The duration of one set of measurements was 10 min. The interval between subsequent sets of measurements was 60 min.

The mean O_2cal for one set of measurements (10-min period) was calculated from the mean values of O_2cal obtained per minute. Simultaneously, the mean value for continuously measured cardiac output in 30 patients, with a 10 min period) for each set of measurements, was calculated from the mean values measured during a 1-min period.

Data for CO obtained from 45 patients through the conventional method were calculated as the mean of three triplicate measurements made simultaneously with measurement of O_2cal over a 10-min period.

Immediately after each determination of CO, blood was slowly drawn after aspiration of the catheter dead space from the distal lumen of the pulmonary artery catheter and simultaneously from the arterial line. Blood gas analysis was performed without delay.

Lactate and Glucose Metabolism

Directly after samples had been taken for calculation of O₂ content, arterial and mixed venous blood for lactate and glucose determination were sampled. The transpulmonary lactate gradient was calculated as the difference between the mixed venous and arterial lactate concentrations. Lactate flux was defined as the product of the lactate gradient and CO. The transpulmonary glucose gradient was calculated as the difference between the mixed venous and the arterial glucose concentrations. Glucose flux was defined as the product of the glucose gradient and CO.

BAL

BAL was done once daily on each patient, immediately after O₂ measurement. A bronchoscope was introduced through the endotracheal tube during continuing mechanical ventilation. The distal end of the bronchoscope was gently wedged into the lateral segment of the right middle lobe, and 150 ml of sterile 0.9% NaCl at room temperature were instilled through the bronchoscope channel and aspirated by applying gentle suction through a syringe. In patients with pneumonia, BAL was performed in a subsegment of the lobe showing infiltration. Total cell counts of lavage fluid were done through flow cytometry. Differential counts were determined by manual counting under the microscope.

Lavage results (cellular analyses) were considered abnormal if they differed significantly from the quality criteria published by Chamberlain and coworkers (14) or from the reference standard of our laboratory.

Statistics

Descriptive statistical parameters were used, and results are expressed as means ± SEM. Reproducibility was assessed from the coefficient of variation (CV). The CVs for intrapatient O_2pulm measurements were calculated on each day, as well as between days, over the 3-d measurement protocol. Linear regression analysis was used to assess the relationship between changes in O_2pulm and the variables of total O₂ consumption, (A-a)DO₂, venous admixture, PEEP, static compliance, transpulmonary lactate flux, and transpulmonary glucose flux. In addition, the relationships between O_2Fick and O_2cal, between lactate flux and mixed venous lactate concentration, and between glucose flux and mixed venous glucose concentration were determined by means of linear regression analysis. Statistically significant differences between groups were determined with either the Mann-Whitney rank-sum test or ANOVA when appropriate. A value of p < 0.05 was considered statistically significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Demographic data, clinical characteristics, and global parameters of oxygen transport are shown for all patients in Table 1.

In 41 patients with pneumonia, 286 serial measurements were made, and in 34 patients without pneumonia 255 such measurements were made. Criteria for healed pneumonia were fulfilled in 24 patients (148 serial measurements), on an average of 10 d (range: 5 to 21 d) after the first evidence of pulmonary infiltration was seen. In all patients with pneumonia, pulmonary infiltrates and purulent tracheal secretions were found. Radiographic proof of pulmonary infiltrates was obtained for all patients from chest X-rays and was confirmed in 19 patients by CT scan. In addition, a diagnosis of pneumonia was associated in 28 cases with quantitative culture of lower respiratory tract secretions, in 15 cases with positive blood culture of an organism found in lower respiratory tract secretions, in seven cases with positive pleural fluid culture of an organism found in lower respiratory tract secretions, and in three cases with histologic proof of VAP on autopsy.

Cardiac index (with pneumonia: 4.2 ± 0.3 L/min/m²; without pneumonia: 4.3 ± 0.2 L/min/m²), mixed venous oxygen saturation (with pneumonia: 72 ± 1%; without pneumonia: 71 ± 1%), and hemoglobin content (with pneumonia: 10.9 ± 1.2 g/ dl; without pneumonia: 11.2 ± 1.3 g/dl) showed no significant differences between the two study groups.

The magnitude of O_2pulm was independent of the method used to determine CO. No significant differences were seen between values determined by continuous CO measurement or conventional CO measurement.

Figure 1 shows the relationship between simultaneous O_2Fick and O_2cal measurements for all patients. The corresponding results for O_2pulm, apportioned into patients with pneumonia, without pneumonia, and with healed pneumonia, are summarized in Figure 2. As compared with O_2pulm in patients without pneumonia (19.4 ± 1.2 ml/min/m²) and in patients with healed pneumonia (20.3 ± 1.4 ml/min/m²), O_2pulm was significantly increased (p < 0.001) in patients with acute pulmonary inflammation (50.7 ± 1.7 ml/min/m²). As a portion of total oxygen consumption, was O_2pulm 25 ± 1% in the pneumonia-group versus 11 ± 0.5% in the nonpneumonia group (p < 0.001). In all patients there was a significant relationship between O_2pulm and the total O₂ consumption rate (R = 0.58; p < 0.001). Pneumonia was unilobar in 15 patients (O_2pulm: 48.6 ± 1.3 ml/min/m²), bilobar in 16 patients (O_2pulm: 49.1 ± 1.6 ml/min/m²), and multilobar (three or more infected lobes) in 10 patients (O_2pulm: 62.0 ± 1.4 ml/min/m²; p < 0.001, compared with patients with unilobar as well as bilobar pneumonia). The relationship between O_2pulm and various other parameters of lung function is shown in Table 2.

View larger version (26K): [in this window] [in a new window]   Figure 1.   Correlation by linear regression analysis between simultaneous O2Fick and O2cal measurements in patients with pneumonia (286 sets of measurements) and in patients without pneumonia (255 sets of measurements).

View larger version (12K): [in this window] [in a new window]   Figure 2.   Pulmonary oxygen consumption ( O2pulm). A: Patients with pneumonia (286 sets of measurements). B: Patients without pneumonia (255 sets of measurements), C: Patients with healed pneumonia (148 sets of measurements) (*significant difference from Group B and C. Cross-lines symbolize the mean value for each group).

On each day, the intrapatient CV for O_2pulm measurements was 3.2 ± 0.2%. Between days (over the 3-d measurement protocol), the intrapatient CV for O_2pulm measurements was 6.1 ± 0.3%.

In five patients with pneumonia and in 11 patients without pneumonia, an accumulation of interstitial fluid within the lung was detected by chest radiography. The subsequently measured EVLW ranged from 8.5 ml/kg to 18.7 ml/kg. There was no significant correlation between EVLW and O_2pulm.

The recovery rate of instilled bronchoalveolar lavage fluid (BALF) was approximately 50 to 60% from the right middle lobe as well as from other pulmonary regions. BALF cellularity (51 ± 12 cells × 10⁴/ml in patients with pneumonia, versus 34 ± 9 cells × 10⁴/ml in patients without pneumonia; p < 0.01), percent macrophages (18 ± 6% in patients with pneumonia, versus 54 ± 10% in patients without pneumonia; p < 0.01), and percent neutrophils (73 ± 9% in patients with pneumonia, versus 38 ± 8% in patients without pneumonia; p < 0.01) were significantly different in the two groups.

Lactate levels were significantly higher in mixed venous blood (1.95 ± 0.1 mmol/L in patients with pneumonia versus 1.62 ± 0.1 mmol/L in patients without pneumonia) than in the arterial blood (1.75 ± 0.1 mmol/L in patients with pneumonia versus 1.5 ± 0.1 mmol/L in patients without pneumonia) in both groups (p < 0.001 for both comparisons). The transpulmonary lactate gradient was 0.19 ± 0.02 in the pneumonia group versus 0.12 ± 0.02 in the group without pneumonia (p < 0.001). The relationship between O_2pulm and transpulmonary lactate flux is shown for both groups in Figure 3. Furthermore, there was a significant relationship between mixed venous lactate concentration and transpulmonary lactate flux (R = 0.24; p < 0.05).

View larger version (17K): [in this window] [in a new window]   Figure 3.   Relationship by linear regression analysis between O2pulm and transpulmonary lactate flux in patients with pneumonia (286 sets of measurements) and in patients without pneumonia (255 sets of measurements).

Whereas in patients with pneumonia the arterial glucose concentration (7.1 ± 0.5 mmol/L) was significantly lower than the mixed venous glucose concentration (7.5 ± 0.6 mmol/L; p < 0.01), there was no significant difference between arterial (7.2 ± 0.7 mmol/L) and mixed venous (7.3 ± 0.6 mmol/L) glucose levels in patients without pneumonia. No significant correlation was found between mixed venous glucose concentration and transpulmonary glucose flux. The relationship between O_2pulm and transpulmonary glucose flux is shown for both groups in Figure 4.

View larger version (20K): [in this window] [in a new window]   Figure 4.   Relationship by linear regression analysis between O2pulm and transpulmonary glucose flux in patients with pneumonia (286 sets of measurements), and in patients without pneumonia (255 sets of measurements).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The lungs are normally considered to be gas-exchange organs with little energy requirements for themselves (15). Under pathologic conditions, aerobic as well as anaerobic processes consuming oxygen may increase (16). Recently, the O_2pulm was determined directly in patients undergoing total cardiopulmonary bypass surgery by separation of the lungs from pulmonary arterial blood flow (15). Under those conditions, a O_2pulm of about 11 ml/min was estimated. Since it is difficult to assess metabolic requirements of the lungs, indirect approaches, such as the method in the present study, have been used (1, 2).

However, because of the intrinsic error in the measurement of blood gases, hemoglobin, oxygen saturation, and CO necessary to calculate O_2Fick, this approach to determining O_2pulm has been controversial (17). In our study we standardized the techniques mentioned earlier for these measurements as much as possible.

In addition, an increase in incorrect O_2cal measurements at high FI_O2 levels has been reported (18, 19). Such measurement error can be minimized by using a differential paramagnetic oxygen sensor so that the results remain accurate up to an FI_O2 level of 75% (20). In the majority of our patients (88%), an FI_O2 of less than 60% was applied.

Hitherto, six studies dealing with O_2pulm in patients with acute respiratory failure from various causes have been reported (1, 2, 21). The reported data differ greatly in terms of the number of examined patients studied, the underlying disease, the severity of illness, the pattern of ventilation, and even the pulmonary pathology. The mean O_2pulm ranged from 18 to 89 ml/min/m².

In our study, pneumonia was associated with a moderate increase in the severity of illness as assessed by the APACHE II score. The severity of lung injury as assessed from the Murray score showed distinctly higher values in the pneumonia group than in the nonpneumonia group. This difference was caused by alterations in both respiratory mechanics and pulmonary gas exchange, and was reflected in a lower lung compliance, a higher radiographic score, higher levels of PEEP, and a lower oxygenation index. Disturbances in gas exchange capacity were seen in all patients with pneumonia, but also in some patients without pneumonia. In the case of the nonpneumonia group this phenomenon was mainly due to accumulation of ELVW.

Two sets of data have been reported for patients with bacterial pneumonia (4, 17). Whereas Becq and coworkers described an increased O_2pulm in patients with pneumonia, Weyland and colleagues found no significant increase in O_2pulm in cases of pulmonary infection. Weyland and colleagues concluded that because of the poor reproducibility of O_2Fick measurements, the use of method comparison studies for estimating O_2pulm was limited. In the present study, the CV for intrapatient measurements of O_2pulm was less than 4% for each study day. Thus, a sufficient reproducibility for biologic systems was achieved in our study. The higher CV between study days was due to fluctuations in patients' conditions from day to day. That O_2pulm increased with the number of afflicted lobes in our patients seems to confirm that the increased O₂ of infected lungs is of physiologic importance, which is also reflected in the decreased O_2pulm in most of the patients from the same group when they had recovered from pneumonia (healed pneumonia).

However, the question still remains of whether the considerably increased O_2pulm infected lungs is caused by intra- or extrapulmonary structures. At least two physiologic explanations for a between-method difference in O₂ determination using the reverse Fick method and indirect calorimetry must be considered: First, coronary blood flow returning directly to the left side of the heart rather than to the right, via the endocardial thebesian veins, represents a portion of cardiac oxygen consumption that would not be detected by the reverse Fick method. A second possibility is an increase in bronchial blood flow. Like thebesian blood flow, desaturated bronchial flow returning to the left atrium via the pulmonary veins would not be detected by the reverse Fick method.

Under pathologic conditions, as in the case of pneumonia, increased O_2pulm has been interpreted as a result of enhanced metabolic activity of essentially extrapulmonary structures such as neutrophils or macrophages, which are well equipped with mitochondria (6, 25). It is known that these cells proliferate during pulmonary infection and are apparent in BALF (1). In our study, BALF cellularity and the percentage of neutrophils increased significantly in patients with pneumonia. This observation seems to confirm the involvement of inflammatory cells such as neutrophils and macrophages generating reactive oxygen species, potentially contributing not only to aerobic but also to anaerobic lung metabolic activity. We speculate that these cells compete with parenchymal cells of the lung for the available amount of oxygen, which may result in an oxygen steal from intrapulmonary structures.

In addition to an altered pattern of pulmonary blood flow and increased cellular oxygen consumption by neutrophils and macrophages within the lungs, acute lung injury may cause changes in the metabolic function of lung parenchymal cells such as pneumocytes and pulmonary vascular endothelial cells (26). It could be argued that an increase in oxygen-dependent metabolic processes is necessary because the lungs have little energy reserve and are highly dependent on circulating substrates. With regard to the difference between arterial and mixed venous lactate concentrations, our results are in contrast to those of previous studies. It was suggested by others that the lung is a source of lactate release in critically ill patients with hyperlactatemia, particularly in cases of acute lung injury (26). Although positive differences in arterial- and mixed venous lactate concentration were reported repeatedly, the etiology of lung lactate release remains unclear. It was supposed that this phenomenon is due to the injury of pulmonary vascular endothelial cells. Alternatively, inhibition of pyruvate dehydrogenase, mainly in septic patients, a decrease in lung adenosine triphosphate, or sequestration of neutrophils in the pulmonary circulation has been postulated (26). As in the study by Lee and colleagues (27), lung injury in our patients did not appear to have caused lung lactate release. Moreover, because of the negative difference in arterial-mixed venous lactate concentration in patients with as well as without pneumonia, our results suggest that lactate metabolization by lung tissues may be due to an increase in lactate concentration in mixed venous blood. However, the highly significant relationship between transpulmonary lactate flux and O_2pulm in the pneumonia group indicates that pneumonia is the main factor for increasing lung lactate utilization, mobilizing even the most unlikely resources to meet the energy requirement of infected tissues.

To interpret the differences between our results and those of others regarding the release or utilization of lung lactate, several factors have to be taken into consideration. One possible interpretation was supplied by the earlier findings of Strauss and associates (28) and Caldwell and coworkers (29), to the effect that consolidated lungs with granulomatous disease produced lactate under hypoxic conditions or metabolized lactate under normoxic conditions. Because in most of our patients increased lactate utilization was found, hypoxia of the lung parenchyma seems to be unlikely, particularly since the lungs are supplied by three sources of oxygen: the alveolar gas; the mixed venous blood which continues to carry oxygen; and the bronchial circulation. Similarly, in the pneumonia group, glucose flux was increased, indicating increased glucose utilization.

Increased glycogen synthesis, using lactate and glucose in particular as substrates, has been reported in fetal lungs (30). Increased requirements for glycogen can be caused among other things by the production of surfactant phospholipids, which can be impaired in acute lung injury. In adult as opposed to fetal lungs (30), no increase in glycogen synthesis has so far been reported. Nevertheless, the hypothesis that in infected lung tissue the pattern of substrate utilization and capacity of lung cells for oxygen uptake and metabolism is altered should be investigated in further studies.

In our patients no correlation between the level of O_2pulm and mortality was found. Whether an increased O_2pulm has any prognostic value is open to question.

In conclusion, our results show that at least in the case of pulmonary infection, the difference between O_2cal and O_2Fick appropriately reflects O_2pulm. Furthermore, there seems to be a correlation between the extent of pulmonary inflammation and the magnitude of O_2pulm, as reflected in the increase in O_2pulm with an increased number of affected lobes. Activated inflammatory cells such as neutrophils and macrophages might be involved in lung metabolic processes consuming oxygen. The hypothesis that the increased O_2pulm in pneumonia may be caused at least in part by an increase in oxygen-dependent metabolic processes within the infected lung leaves room for further investigation. A practical pathophysiologic consequence of our findings is that O_2pulm in pneumonia is an essential component of up to 30% of total O₂, and the magnitude of O_2pulm might significantly affect arterial oxygenation. This effect may cause an overestimation of S/T measured with the O₂ method and an underestimation of O₂ as measured with the reverse Fick method.