INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Mechanical ventilation allows patients to recover from acute respiratory failure and major surgical procedures (1). Discontinuation of ventilator support, however, is difficult in about a third of patients (2, 3). Management of these patients is particularly difficult, largely because of our limited understanding of the pathophysiologic mechanisms responsible for weaning failure. We have recently shown that inspiratory muscle effort is markedly increased in patients who fail a weaning trial (4). The associated increase in intrathoracic pressure excursions may lead to complex cardiopulmonary interactions. Indeed, Lemaire and coworkers (5) documented sudden increases in pulmonary artery occlusion pressure (Ppao) during unsuccessful weaning attempts in patients who had both chronic obstructive pulmonary disease (COPD) and cardiovascular disease; patients without cardiac disease were not included in the study.

To better understand the importance of hemodynamic performance as a determinant of weaning outcome, we obtained measurements in ventilator-supported patients who failed a trial of spontaneous breathing and in a control group who tolerated the trial and were extubated successfully. The rapid and progressive development of cardiopulmonary failure in patients failing a weaning trial (4) may be characterized incompletely by many tests of cardiovascular performance employed in a critical care setting since they provide only intermittent assessment. Accordingly, we employed a pulmonary artery catheter that provides continuous measurements of mixed venous oxygen saturation (Sv_O2) as our primary tool. Supplementary data included intermittent measurements of cardiac output and pulmonary and systemic vascular pressures. We hypothesized that patients who fail a trial of spontaneous breathing develop a decrease in Sv_O2 because of an increase in the O₂ cost of breathing, which is met by insufficient transport of O₂ to the tissues.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

Nineteen medical and surgical patients (16 men, 3 women 66 ± 2 SE yr of age) who were hemodynamically stable and whose primary physician considered them ready to undergo a trial of weaning were recruited (Table 1). The patients were ventilated in the assist-control mode using a Servo 900C (Siemens, Schaumburg, IL) or Puritan-Bennett 7200a (Puritan-Bennett, Los Angeles, CA) ventilator through a cuffed endotracheal tube. All patients had systemic and pulmonary artery catheters (Oximetrix; Abbott Laboratories, North Chicago, IL) inserted as part of their medical management. Waveform characteristics, blood gas sampling, and chest radiographs before the study confirmed satisfactory positioning of each pulmonary artery catheter. The study was approved by institutional ethics committees, and informed consent was obtained from each patient.

Hemodynamic Measurements

An indwelling arterial line and the distal and proximal ports of the pulmonary artery catheter were connected to a strain-gauge manometer that provided continuous recordings of mean systemic arterial, pulmonary arterial, and right atrial pressures, respectively. The Ppao was determined after balloon inflation and read at end-expiration. Cardiac output (T) was measured by the thermodilution principle using 10-ml aliquots of saline at room temperature randomly injected throughout the respiratory cycle, and the average of five measurements was taken.

Gas Exchange

Oxygen tension (PO₂), carbon dioxide tension (PCO₂), pH, oxyhemoglobin saturation (SO₂), and hemoglobin concentration were determined from blood samples drawn anaerobically through the arterial line and the distal port of the pulmonary artery catheter. The samples were analyzed immediately with an IL system 1303 pH/blood gas analyzer (Instrumentation Laboratories, Lexington, MA) and a Model IL 282 CO-oximeter (Instrumentation Laboratories).

Continuous recordings of Sv_O2 were sampled by a system that included the pulmonary artery catheter, optical module, and computer (Oximetrix; Abbott Laboratories, Chicago, IL) (7). Before insertion, the light intensity of each catheter was verified using a reflection standard. After insertion, Sv_O2 measured by the catheter was calibrated against a direct photometric measurement (CO-oximeter) using an aspirated sample of mixed venous blood; if the two measurements differed by > 4%, the system was recalibrated to display the photometric value of Sv_O2. Artifacts resulting from impingement of the catheter tip on a vessel wall or formation of a clot over the optics were avoided by monitoring the intensity of the reflected light. Data were digitized and stored on computer disk for subsequent data analysis.

Protocol

Each patient was receiving assist-control ventilation at the time data collection commenced. Recordings were first obtained during 10 min of mechanical ventilation, and a trial of spontaneous breathing was then initiated. The patient was placed in a semirecumbent position and breathed through a T-tube circuit, receiving the same fractional inspired oxygen concentration as during mechanical ventilation. A priori criteria for weaning failure were frequency > 35 breaths/min, arterial SO₂ < 90% on pulse oximetry, heart rate > 140 beats/min or a sustained increase/decrease in the heart rate of more than 20%, systolic arterial pressure above 180 mm Hg or below 90 mm Hg, increased accessory muscle activity, diaphoresis and facial signs of distress (2). Patients who met these criteria were returned to mechanical ventilation and designated the failure group. Patients free of these features at the end of the trial were extubated and designated the success group. Continuous measurements of Sv_O2, systemic arterial, pulmonary arterial, and right atrial pressures were monitored during mechanical ventilation and throughout the trial. Cardiac output, and arterial and mixed venous blood gas measurements were measured while the patient received mechanical ventilation and at 5, 15, 30, 45, and 60 min after start of the trial of spontaneous breathing, depending on its duration.

Physiological Measurements

Systemic arterial O₂ content (Ca_O2, ml/dl) was calculated as Ca_O2 = 1.34 × (Hb × Sa_O2/100) + (0.003 × Pa_O2); mixed venous O₂ content (Cv_O2) was calculated in the same manner using Sv_O2 and Pv_O2; O₂ transport (O₂, ml/min/kg) was calculated as T × Ca_O2; O₂ consumption (O₂, ml/min/kg) was calculated as T × (Ca_O2  Cv_O2); and O₂ extraction ratio as O₂/O₂. Venous admixture (VA/T) was calculated as Cc'_O2  Ca_O2/Cc'_O2  Cv_O2, in which Cc'_O2 represents the O₂ content of "ideal" end-capillary blood derived from alveolar PO₂, as estimated by the alveolar gas equation.

Data Analysis

Physiological variables in the successful and unsuccessful outcome groups were compared during mechanical ventilation by unpaired t tests. Because the trial duration varied among patients, the continuous Sv_O2 data in each patient were re-sampled at intervals of 4% of trial duration, with zero being the point at which the ventilator was discontinued and 100% representing the end of the spontaneous breathing trial. The resulting 25 equal time intervals from each patient were aligned and superimposed with respect to time, and an ensemble average was calculated for each group at a given time point. Hemodynamic and gas-exchange data collected during the fifth and last minute of the trial were compared by one-way analysis of variance with repeated measures; the Newman-Keuls test of multiple comparisons was used to assess differences between individual means when appropriate. Because of the short duration of the trial in two patients, hemodynamic and gas-exchange data were obtained at only a single time point, and these subjects are not included in analysis of the intermittent data. Data between the groups were compared by two-way analysis of variance with repeated measures across time. Results are expressed as mean ± standard error (SE).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

During the trial of spontaneous breathing, eight patients met the a priori criteria for weaning failure after 39 ± 9 min and mechanical ventilation was reinstituted. Eleven patients tolerated the trial without distress and were extubated after 43 ± 2 min.

Mixed Venous Oxygen Saturation

During mechanical ventilation, Sv_O2 was not significantly different between the failure and the success groups (Figure 1). At the moment that ventilator support was discontinued, Sv_O2 in the failure and success groups was 61.3 ± 5.8 and 65.4 ± 1.6%, respectively. At the end of the trial, Sv_O2 decreased to 51.5 ± 7.9% in the failure group (p < 0.01), whereas it remained unchanged in the success group. Over the course of the trial, Sv_O2 was lower in the failure group than in the success group (p < 0.02).

View larger version (28K): [in this window] [in a new window]   Figure 1.   Ensemble averages of the interpolated values of mixed venous oxygen saturation (SvO2) during mechanical ventilation and a trial of spontaneous breathing in the success group (open symbols) and failure group (closed symbols). During mechanical ventilation, SvO2 was similar in the two groups (p = 0.28). Between the onset and the end of the trial, SvO2 decreased in the failure group (p < 0.01) whereas it remained unchanged in the success group (p = 0.48). Over the course of the trial, SvO2 was lower in the failure group than in the success group (p < 0.02). (Bars represent SE.)

Arterial Oxygenation and Oxygen Transport

During mechanical ventilation, Sa_O2 was lower in the failure than in the success group (p < 0.005), and it remained lower throughout the trial (Table 2). Venous admixture (VA/T) tended to be higher in the failure group during mechanical ventilation (p = 0.07), and over the course of the trial it was significantly higher (p < 0.0006).

Three interrelated phenomena, O₂, O₂, and O₂ extraction ratio, are shown in Figure 2. O₂ was not significantly different in the failure and success groups during mechanical ventilation or at the onset of the trial of spontaneous breathing. Compared with the value during mechanical ventilation, O₂ tended to increase at the end of the trial in the success group (p = 0.08), whereas it did not change in the failure group. Over the course of the weaning trial, O₂ differed in the two groups (p < 0.02) and changed in opposite directions (Figure 2). O₂ was similar in the two groups during mechanical ventilation and at the onset of the weaning trial, and it changed in neither group over the course of the trial. During mechanical ventilation, O₂ extraction ratio did not differ significantly between the two groups (p = 0.40). On discontinuation of mechanical ventilation, O₂ extraction ratio immediately decreased in the success group (p < 0.05), but not in the failure group; by the end of the weaning trial, the proportional change in O₂ extraction ratio from the value during mechanical ventilation was greater in the failure group than in the success group (p < 0.02). In summary, during the weaning trial the failure group had a relatively lower O₂ and a higher O₂ extraction ratio than did the success group. Parenthetically, Sv_O2 in the failure group was strongly correlated with O₂ (r = 0.78, p < 0.0001) and with O₂ extraction ratio (r = 0.83, p < 0.0001); the relationship between Sv_O2 and O₂ was not significant (r = 0.36, p = 0.12) (Figure 3). It is important to note that the correlations between Sv_O2 and O₂ and between Sv_O2 and extraction ratio do not involve mathematical coupling (8).

View larger version (18K): [in this window] [in a new window]   Figure 2.   Oxygen transport, oxygen consumption, and isopleths of oxygen extraction ratio in the success (open symbols) and failure (closed symbols) groups during mechanical ventilation (squares) and at the onset (circles) and end (triangles) of a spontaneous breathing trial. See text for details.

View larger version (16K): [in this window] [in a new window]   Figure 3.   The relationships between mixed venous oxygen saturation (SvO2) and O2 transport (upper left panel ), O2 extraction ratio (upper right panel ), and O2 consumption ( O2) (lower panel ) in the failure group. SvO2 was strongly correlated with O2 (r = 0.78, p < 0.0001) and with O2 extraction ratio (r = 0.83, p < 0.0001); the relationship between SvO2 and O2 was not significant (r = 0.36, p = 0.12).

To determine whether the increase in O₂ extraction was due to an increase in anaerobic metabolism of the respiratory muscles, we estimated the value of pH at the end of the trial by assuming that the change in pH was solely due to an increase in PCO₂. We then compared this predicted value of pH with the actual measured value (Figure 4). The predicted and actual values of pH were closely related to each other in both the failure group (r = 0.99, p < 0.0001) and the success group (r = 0.83, p < 0.002), suggesting that the observed acidosis was respiratory rather than metabolic in nature.

View larger version (11K): [in this window] [in a new window]   Figure 4.   Measured pH at the end of the trial of spontaneous breathing compared with values predicted for a situation where the change in pH can be explained solely by an increase in PaCO2. In both the failure (closed symbols) and success groups (open symbols), the points lie close to the line of identity, indicating that the observed acidosis was of respiratory rather than metabolic origin.

Hemodynamic Variables

During mechanical ventilation, cardiac index was similar in the failure and success groups (Table 3). At the end of the trial, cardiac index increased in the success group (p < 0.05), but they did not change from mechanical ventilation in the failure group. During mechanical ventilation, mean pulmonary artery pressure () was higher in the failure group than in the success group (p < 0.009), whereas right atrial pressure (PRA), Ppao, and mean arterial pressure () were similar in the two groups (Table 3). On discontinuation of the ventilator, PRA, , Ppao, and increased in the failure group, but they did not change in the success group. Over the course of the trial, PRA, , Ppao, and were higher in the failure group than in the success group (p < 0.025 in each instance). The failure group had equivalent values of cardiac index during mechanical ventilation and at the end of the weaning trial, whereas was significantly higher by the end of the trial (p < 0.05), suggesting an increase in right ventricular afterload (Figure 5). Likewise, the increase in from the value during mechanical ventilation to that at the end of the unsuccessful trial, together with the lack of change in cardiac index, suggests an increase in left ventricular afterload (Figure 5).

View larger version (14K): [in this window] [in a new window]   Figure 5.   Mean pulmonary artery pressure () and mean arterial pressure () versus cardiac index (CI) during mechanical ventilation (squares) and at the onset (circles) and end (triangles) of a spontaneous breathing trial in the success (open symbols) and failure group (closed symbols). The shaded area represents the normal range of increase in with CI (35). In the success group, CI increased between mechanical ventilation and the end of the trial, remained slightly above the normal range, and remained unchanged. In contrast, in the failure group, CI was similar during mechanical ventilation and at the end of the weaning trial, but and were higher by the end of the trial (p < 0.025 and p < 0.05, respectively). The increases in and , together with the lack of change in CI, strongly suggest increases in right- and left-ventricular afterload in the failure group.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Immediately before a trial of spontaneous breathing, the values of S_O2 in patients who went on to fail the trial were not significantly different from those in a control group who tolerated the trial and were successfully extubated. On discontinuation of mechanical ventilation, S_O2 fell progressively in the failure group, whereas it remained unchanged in the success group. Over the course of the trial, O₂ increased in the success group. whereas it did not change in the failure group. The two groups had similar O₂ demand, which the failure group achieved by greater O₂ extraction, resulting in a decrease in S_O2.

Arterial Oxygenation

Impairment in arterial oxygenation may have contributed to the decrease in S_O2 in the failure group. The low Sa_O2 was partly due to the higher VA/T in the failure group, as suggested by the correlation between the variables (r = 0.74, p = 0.0005). The increase in VA/T with resumption of spontaneous breathing in the failure patients is probably due to derecruitment of lung units secondary to withdrawal of positive-pressure ventilation (9). Another mechanism contributing to arterial desaturation is the reduction in S_O2 (10). In a study of patients with COPD who developed hypoxemia during low-level exercise, almost all of the decrease in Pa_O2 could be explained by a fall in P_O2 (11).

O₂ Transport

The different responses in S_O2 during the trials of spontaneous breathing in the two groups can be largely explained by the opposite changes in O₂further supported by the correlation between O₂ and S_O2 (r = 0.78, p < 0.0001) (Figure 3). The difference in O₂ responses is mostly due to changes in cardiac index, which increased in the success group but did not change in the failure group. An increase in cardiac index is the anticipated response to discontinuation of mechanical ventilation in patients with intact cardiovascular function (10) and is largely mediated by an increase in preload (12, 13). The apparent lack of increase in PRA on discontinuation of mechanical ventilation in the success group may have occurred because ours was a non-transmural measurement. Transmural pressure is calculated by subtracting pleural pressure from the hemodynamic pressure, and since intrathoracic pressure is predominantly negative during spontaneous breathing, transmural PRA will be greater than nontransmural PRA (12).

The relative decrease in cardiac index in the failure group could have resulted from several mechanisms. One, a decrease in preload, but this possibility seems unlikely since PRA increased in the failure group; however, an increase in alveolar pressure secondary to expiratory muscle contraction (14) may have contributed to the increase in PRA. Two, a decrease in cardiac index could have resulted from impairment in cardiac contractility consequent to respiratory acidosis and hypoxemia (15, 16). An impairment in cardiac contractility could also result from myocardial ischemia, which has been documented during weaning by some (17, 18) but not other (19) investigators; however, impairment of left ventricular contractility because of the development of myocardial ischemia during weaning appears to be largely confined to patients with documented coronary artery disease (19, 20). Three, an increase in left ventricular afterload, because of more negative swings in intrathoracic pressure (4, 21), probably contributed to the decrease in cardiac index (Figure 5). An increase in left ventricular afterload is probably also responsible for the increase in Ppao in the failure group over the course of the weaning trialalthough the contribution of expiratory muscle contraction cannot be excluded.

O₂ Consumption

An increase in O₂ demand during the trial of spontaneous breathing could contribute to a decrease in S_O2, but this does not appear to have been the case since O₂ did not change in either patient group. That a change in O₂ demand was not responsible for the decrease in S_O2 in the failure patients is further supported by the poor correlation between the two variables (r = 0.36) (Figure 3). However, three patients in the failure group showed individual increases in O₂ of 21, 24, and 69% between the onset and end of the weaning trial, most, if not all, of which can be attributed to an increase in the O₂ cost of breathing (22).

O₂ Extraction

When O₂ is plotted against a wide variation in O₂, a biphasic relationship is observed (23, 24). Above a critical level of O₂, a change in O₂ is met by a proportional and inverse change in O₂ extraction so that O₂ remains constant. Below this critical threshold, O₂ is linearly dependent on O₂, and O₂ extraction is unable to compensate for a reduction in O₂. The critical threshold for O₂ that separates the dependent and independent portions of the O₂-O₂ relationship is thought to be 4.5 ml/min/kg (range, 2.8 to 6.2 ml/min/kg) (24). At the end of the trial of spontaneous breathing, our weaning failure patients had a O₂ of 10.8 ml/min/kg, which is considerably above the proposed critical valuealthough one patient had a O₂ of 6.4 ml/min/kg. Oxygen extraction is normally 0.20 to 0.30 (25) and it increases to greater than 0.80 during maximal exercise (26), probably as a result of improved O₂ diffusion secondary to an increase in tissue capillary density (27). When O₂ extraction is less than 0.60 increased metabolic demand appears to be met aerobically, whereas energy requirements are met by anaerobic pathways when O₂ extraction exceeds 0.60 (28). Between mechanical ventilation and the end of the trial of spontaneous breathing, O₂ extraction ratio increased in the failure group; the value at the end of the trial (0.37) did not reach the ratio reported to signify the onset of anaerobic metabolism (0.60), although two patients had ratios of 0.50 and 0.56. Moreover, the respiratory origin of the acidosis in the failure patients (Figure 4) suggests that energy requirements were met solely by aerobic pathways.

The ability of the failure group to deal with respiratory muscle energy demands through aerobic pathways is probably related to the capacity of the diaphragm to achieve higher blood flow than most other skeletal muscles (29). During loading, the diaphragm rapidly increases O₂ extraction to a plateau of ~ 55 to 65%; further increases in O₂ are achieved by increases in blood flow (30, 31). The diaphragmatic musculature appears to be extremely resistant to hypoxic stress, and animals can maintain a ventilation that is sufficient to avoid hypercapnia until phrenic vein PO₂ falls to 12 mm Hg (29). Likewise, investigators have found that a PO₂ of ~ 10 mm Hg in the phrenic vein is the threshold associated with the onset of diaphragmatic lactate production (32) and the development of fatigue (33). The lowest P_O2 in our patients was 26 mm Hg, which is above the threshold for onset of diaphragmatic lactate production. This association needs to be interpreted with caution, however, since mixed venous blood contains effluents from many tissue beds other than the diaphragm. To our knowledge, this is the first documentation that when challenged by an increase in mechanical load (4), the respiratory muscles of critically ill patients do not appear to switch from aerobic to anaerobic metabolism.

Pulmonary Hypertension

was higher in the failure group than in the success group during mechanical ventilation and it increased further at the end of the trial, although calculated pulmonary vascular resistance did not increase. Caution is required, however, when interpreting calculations of pulmonary vascular resistance since they do not necessarily reflect active changes in vascular caliber unless passive mechanisms are taken into account (34). This has led to use of pressure/flow plots, as shown in Figure 5. Cardiac index values were equivalent at the end of the trial and during mechanical ventilation, although was significantly increased at the former point, which strongly suggests an increase in right ventricular afterload. Moreover, the pulmonary artery pressures in the failure group were more than twice the values observed in the weaning success patients (Figure 5). The increase in in the weaning failure patients is probably multifactorial in origin. Hypoxemia and acidosis, both of which occurred in our failure patients, are potent vasoconstrictors. In addition, pulmonary artery pressure can be increased by alveolar vessel compression (12, 20) because of the increase in alveolar pressure that accompanies the dynamic hyperinflation and deterioration in pulmonary mechanics in patients who fail a weaning trial (4, 6).

In summary, patients who failed a trial of spontaneous breathing had similar S_O2 values immediately before the trial as a control group who were successfully extubated. Upon discontinuation of ventilator support, a progressive decrease in S_O2 was noted in the failure group but not in the success group. During the trial, the failure group behaved differently than the success group, who demonstrated increases in cardiac index and O₂ transport over the values during mechanical ventilation; the lack of increase in convective O₂ transport in the failure group was due, in part, to elevated right- and left-ventricular afterload. Instead, patients in the failure group increased O₂ extraction by the tissues, which combined with a relative decrease in O₂ transport, resulted in a fall in mixed venous oxygen saturation.