INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Weaning from mechanical ventilation is an important consideration in the daily operation of intensive care units because this proceeding can represent more than 40% of mechanical ventilation duration (1). Difficulty in discontinuation of ventilatory support is observed in about 20% of patients receiving mechanical ventilation, and chronic obstructive pulmonary disease (COPD) constitutes an independent factor influencing the length of weaning, and is an issue in weaning strategy (2).

Several pathophysiological determinants of weaning failure have been identified, including inadequacy of pulmonary gas exchange, cardiovascular dysfunction, disturbance in respiratory muscle performance, and psychological factors. Respiratory muscle pump failure, and thus the inadequacy of respiratory muscle pump load and respiratory muscle performance, has been said to be the primary cause of weaning failure (3).

In critically ill patients, and especially those with COPD, the oxygen cost of breathing is increased during the weaning process and may increase the proportion of total body oxygen delivery required by the respiratory muscles to meet ventilation requirements (4, 5). This increased oxygen cost of breathing is met by an increase in blood flow to respiratory muscles, resulting in a blood flow diversion from other tissues, and may lead to hypoperfusion in some areas (6).

Tonometry represents a simple technique for assessing the adequacy of perfusion of the gastrointestinal mucosa (7). Mohsenifar and coworkers have used PCO₂ of gastric juice and calculated intramucosal pH as an early indicator of weaning success or failure during a weaning trial in 29 critically ill patients (8). Patients who could not be weaned from mechanical ventilation had a substantially reduced gastric intramucosal pH during a 30-min weaning trial whereas no change was observed in patients who were successfully weaned from mechanical ventilation (8). Inadequate oxygen delivery to splanchnic organs owing to a reduction of splanchnic blood flow in failed weaning attempts has been proposed as the cause of this gastric intramucosal acidosis (8, 9), but no direct evaluation of gastrointestinal mucosal blood flow has been reported to support this hypothesis.

Laser-Doppler flowmetry permits real-time and continuous monitoring of mucosal gastrointestinal blood flow (10). Although this method does not provide measurement of tissue blood flow in absolute units, it is reliable for estimating relative changes in gastrointestinal mucosal perfusion (11).

The purpose of this study was to determine whether gastric intramucosal pH changes during a mechanical ventilation weaning trial are related to gastric mucosal blood flow modifications in patients with COPD.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

This prospective study was conducted between January 1997 and June 1998 in the Intensive Care Unit of the Calmette University Hospital (Lille, France). The study population included 16 patients with COPD who tolerated a 2-h weaning trial and were successfully extubated and 11 patients with COPD who failed such a trial. All patients were under controlled ventilation for more than 48 h before inclusion. Acute respiratory failure was caused by exacerbation of COPD. The study was approved by our local ethics committee (CP 96/129) and all patients provided written informed consent.

To be enrolled in the study, the patients needed to be considered weanable by their attending physician according to the following criteria: partial or complete recovery from the underlying cause of acute respiratory failure; adequate gas exchange, as indicated by a ratio of the partial pressure of arterial oxygen (Pa_O2) to the fraction of inspired oxygen (FI_O2) above 200 with a positive end-expiratory pressure of less than 5 cm of H₂O; a core temperature less than 38° C; a hemoglobin value greater than 8 g/dl; and no further need for vasoactive and sedative agents. Patients with known left ventricular dysfunction or cardiac arrythmia were excluded.

Protocol

After inclusion, the patients underwent a pressure support (PS) trial, defined as assisted spontaneous breathing with a PS of 8 cm H₂O for 2 h. All patients were studied in a semirecumbent position. The FI_O2 set was kept constant during the PS trial. Minute ventilation was measured with an Evita2 monitoring device (Draeger, Lubeck, Germany) and tidal volume was obtained by dividing the minute ventilation by the respiratory frequency. The evaluation criterion of the trial was weaning success or failure. Trial failure was defined as the presence of one or more of the following criterion during the 2-h PS trial: a respiratory rate of more than 35 breaths/min for 5 min or longer; a heart rate greater than 140 beats/min or a sustained increase or decrease in the heart rate of more than 20%; a systolic blood pressure greater than 180 mm Hg or less than 90 mm Hg; increased anxiety and diaphoresis. Patients who had none of these features throughout the PS trial were extubated. The decision to remove the endotracheal tube was made independently of the study investigators at the end of the PS trial by primary care physicians who did not have access to the specific data obtained. Weaning success was defined as patient tolerance of extubation for more than 24 h after the PS trial. Conversely, weaning failure was declared if extubation could not be performed or if reinstitution of mechanical ventilation was necessary during the first 24 h after extubation.

Measurements

All measurements were recorded under mechanical ventilation just before the PS trial (H₀) and hourly for 2 h (H_{60 min} and H_{120 min}).

Hemodynamic status. Heart rate and mean arterial pressure were recorded at each time of the study protocol. Cardiac output was measured by transthoracic impedance (Bomed NCCOM 3; Bomed Biomedical, Irvine, CA), using a lateral spot electrode configuration and incorporating the Sramek-Bernstein equation (12). The mean of five consecutive determinations of cardiac output was recorded as cardiac output. A satisfactory agreement between this noninvasive method and thermodilution had been observed in critically ill patients under mechanical ventilation (12) and reproducibility is comparable to reference techniques (13).

Gastric mucosal acid-base status. Gastric intramucosal PCO₂ (PCO₂im) was determined with a tonometer (TRIP; Tonometrics, Worcester, MA). All patients received histamine receptor (H2) agents by continuous intravenous infusion (14). Enteral nutrition was stopped 6 h before the beginning of the study protocol. After nasogastric insertion, the position of the balloon in the stomach was checked radiographically before inflation with 2.5 ml of room air temperature normal saline. A 60-min equilibration time was allowed between tonometer saline loading and sampling for measurements. The dead space saline volume of the tonometer balloon (1 ml) was discarded and the remaining volume was anaerobically aspirated for immediate measure of PCO₂. Steady state adjusted PCO₂ was determined with a time-dependent equilibration factor (15). An arterial blood sample was anaerobically obtained simultaneously for determination of arterial pH, blood gases, and bicarbonate concentration (ABL 520; Radiometer, Copenhagen, Denmark). Gastric mucosal acid-base status was assessed by calculation of the intramucosal-arterial PCO₂ difference and intramucosal pH (14). Gastric intramucosal pH (pH_i) was obtained indirectly by substituting in the Henderson-Hasselbalch equation the steady state adjusted mucosal PCO₂ and the arterial blood bicarbonate concentration used as an estimate of gastric tissue bicarbonate (7) obtained at each time of the study protocol.

Gastric mucosal blood flow measurements. Gastric mucosal blood flow (GMBF) was measured continuously with a laser-Doppler flowmeter probe and device (Periflux PF3; Perimed, Jarfalla, Sweden). This technique, as previously described (11), allows real-time and continuous monitoring suitable for gastrointestinal microcirculation inquiries (10). Morever, laser-Doppler flowmetry had been previously validated in animal and human beings by the ⁸⁵Kr technique, by hydrogen and aminopyrine clearance techniques, and by radiolabeled microsphere uptake (16).

The instrument consists of a 2-mW helium-neon laser, a fiberoptic probe, and a photodetector with a signal-processing unit. The fiberoptic probe (PF309), which has an external diameter of 1.7 mm, carries the laser-produced monochromatic beam to and from the gastric mucosa. Instantaneous analysis of the Doppler frequency spectrum and the fraction of back-scattered light that is Doppler shifted corresponds to the product of the mean velocity of red blood cells by the number of moving red blood cells within a certain volume. Hence, the measured quantity is a true estimation of blood flow and not of red blood velocity. Spatial resolution of the optical probe is 1 mm³, and so measurements reflect only the surface where the probe is applied, i.e., the gastric mucosa (17).

Two differents signals are available for external recording. One output is proportional to the intensity of the back-scattered light. A constant back-scattered light of at least 30% of the emitted light indicates an adequate contact of the optical probe with the tissue surface. The second output signal, expressed in volts, is proportional to blood flow and has been shown previously shown to scale linearly with independent measurements of local perfusion in a variety of tissues (18).

Laser-Doppler signal was continuously registered on a personal computer. The gain was adjusted to 1, the cutoff frequency to 12 Hz, and the time constant to 0.2 s. Before each patient was analyzed, a calibration based on the random Brownian motion of small scatterers in an emulsion (Periflux motility standard; Perimed) was performed.

In our population study, laser endoscopic probe was inserted through the inner channel of the nasogastric tube until a stable signal output was obtained. GMBF was then continuously recorded to detect any loss of contact between the probe and the gastric mucosa during the study protocol. Each measurement represented the mean value of a stable GMBF over 90 s and without motion artifacts. To ensure that no gastric air distension happened, and so to avoid the occurrence of any mucosal compression during the weaning attempt, a negative pressure of 20 mm Hg was applied hourly for 1 min after each time of the protocol. Because laser-Doppler mucosal blood flow is expressed in arbitrary units, the measurements of GMBF during each PS trial was expressed as a percentage of the baseline value (H₀).

Statistical Analysis

All data are expressed as means ± SEM. Data were analyzed by analysis of variance (ANOVA) of nonnormalized values for repeated measures on two factors (group and time). Pairwise multiple comparisons were tested by the Student-Newman-Keuls method. Correlation between pooled values of arterial and gastric intramucosal PCO₂ was performed by linear regression analysis. Significance was accepted at p < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

General characteristics including respiratory and gas exchange data, and hemodynamic and gastric mucosal acid-base status of patients during mechanical ventilation, are summarized in Table 1. No significant difference in any of these variables measured at baseline during mechanical ventilation (H₀) was found between patients who were successfully weaned and those who were not.

Pattern of Breathing and Gas Exchange

Respiratory rate increased significantly whereas tidal volume decreased during PS trials in both groups (Figure 1), whereas minute ventilation was unchanged. Significant differences in the pattern of breathing according to the weaning issue were observed, with a lower tidal volume and a higher respiratory frequency in the failure weaning group (Figure 1). This rapid, shallow breathing adopted by the failure weaning group was responsible for acute respiratory acidosis with significant differences in arterial PCO₂ and in arterial pH between the two groups (Table 2). In the same way, arterial PO₂ (Pa_O2) decreased over the 2 h of the PS trial in the failure weaning group (Pa_O2 = 92.2 ± 4 versus 76.1 ± 4 mm Hg for H₀ and H_{120 min}, p < 0.05) while no variation occurred in the success weaning group (Pa_O2 = 80.4 ± 3 versus 84.2 ± 3 mm Hg, for H₀ and H_{120 min}, NS).

View larger version (17K): [in this window] [in a new window]   Figure 1.   Respiratory frequency (breaths · min1) and tidal volume (ml) during a pressure support trial for 2 h in successful weaning group (open columns) and in failure weaning group (solid columns). Data are expressed as means ± SEM. *p < 0.05 versus H0; § p < 0.05 comparing weaning success and failure groups.

Hemodynamic Variables

PS trials over 2 h resulted in an increase in heart rate (HR) for all patients (HR = 89 ± 3 versus 103 ± 3 beats · min¹, for H₀ and H_{120 min}, p < 0.05). Nevertheless, heart rate after 2 h of PS trial was significantly higher in weaning failure patients than in successfully weaned patients (113 ± 5 versus 92 ± 4 beats · min¹, p < 0.05). Mean arterial pressure () also increased in the failure weaning group ( = 92 ± 4 versus 103 ± 4 mm Hg, for H₀ and H_{120 min}, p < 0.05) whereas no change was observed in the success weaning group ( = 98 ± 3 versus 97 ± 3 mm Hg, for H₀ and H_{120 min}, NS). Finally, cardiac index (CI) increased in the failure weaning group at H_{60 min} (CI = 3.3 ± 0.3 versus 3.9 ± 0.3 L · min¹ · m², for H₀ and H_{60 min}, p < 0.05) and at H_{120 min} (CI = 4.0 ± 0.3 L · min¹ · m², for H_{120 min}, p < 0.05 versus H₀) while no change occurred in successfully weaned patients (CI = 2.8 ± 0.2 versus 3.0 ± 0.2 L · min¹ · m²; for H₀ and H_{120 min}, NS). No variation of arterial oxygen delivery index (DO₂) was observed during the PS trial in the failure weaning group (DO₂ = 467 ± 50 versus 523 ± 51 ml · min¹ · m², for H₀ and H_{120 min}, NS) or in the success weaning group (DO₂ = 466 ± 39 versus 492 ± 39 ml · min¹ · m², for H₀ and H_{120 min}, NS).

Gastric Mucosal Acid-Base Status

Gastric intramucosal PCO₂ increased during PS trials in the two groups but was higher at H_{120 min} in the failure weaning group than in the successful weaning group (Table 2). In the same way, gastric intramucosal pH decreased in both groups during the 2 h of the PS trials. However, gastric intramucosal pH was lower in the failure weaning group, with a significant difference at H_{60 min} and H_{120 min} compared with the success weaning group (Table 2). No significant variation or difference in terms of the gastric-arterial PCO₂ difference was observed between the two groups during the PS trial (Table 2). When linear regression was performed between pooled values of regional and systemic PCO₂, a significant linear correlation was found between gastric intramucosal PCO₂ and arterial PCO₂ (r² = 0.70; p < 0.001) (Figure 2).

View larger version (11K): [in this window] [in a new window]   Figure 2.   Relatioship between arterial PCO2 (PaCO2) and gastric intramucosal PCO2 (PCO2im) at H0 (open circles) and at H120 min (closed circles) during a 2-h pressure support trial. r2 = 0.70 for linear regression performed between pooled values of PaCO2 and PCO2im (p < 0.001).

Gastric Mucosal Blood Flow

In the success group, GMBF increased significantly at H_{60 min} (89 ± 36% from baseline; p < 0.05) and at H_{120 min} (85 ± 27% from baseline; p < 0.05) (Figure 3). On the other hand, rapid shallow breathing and weaning failure were associated with a decrease of GMBF at H_{120 min} (22 ± 11% from baseline; p < 0.05) (Figure 3). In addition, when individual GMBF values were indexed by variations of cardiac index, as a ratio between regional and systemic blood flow, the reduction in indexed GMBF in the failure weaning group was more pronounced than GMBF alone (17 ± 12% from H₀ at H_{60 min}; p < 0.05; 35 ± 13% from H₀ at H_{120 min}; p < 0.05).

View larger version (19K): [in this window] [in a new window]   Figure 3.   Cardiac index and gastric mucosal blood flow during a 2-h pressure support trial according to weaning issue (*p < 0.05 versus H0).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The main finding of our study is that although changes in gastric intramucosal pH do occur and differ according to weaning trial issue in patients with COPD, these changes are mainly due to arterial PCO₂ variations rather than gastric mucosal blood flow variations. Nevertheless, GMBF varies according to weaning success or failure: GMBF significantly decreases in patients who developed a weaning failure with respiratory acidosis whereas it increased in patients who were successfully weaned from mechanical ventilation in the same weaning trial.

Weaning Protocol

In our study, pressure support at a level of 8 cm H₂O was selected as a spontaneous breathing weaning protocol. This level has been judged useful to overcome the extra work imposed by breathing through an endotracheal tube and ventilatory circuit (19), although the compensatory levels ranges from 3 to 14 cm H₂O (20). Furthermore, the Spanish Lung Failure Collaborative Group demonstrated that spontaneous breathing trials with PS is as suitable as the T-tube technique for discontinuation of mechanical ventilation (21).

Pattern of Breathing and Hemodynamic Status

The rapid shallow breathing adopted by our failure weaning group is well documented as an early and strong predictor of weaning failure in medical ICU patients (22). This pattern of breathing was responsible for respiratory acidosis in our weaning failure group (Table 2). In 31 patients with COPD undergoing a weaning trial, Jubran and Tobin studied the evolution of respiratory muscle function and respiratory mechanics according to the weaning issue: these authors found that rapid, shallow breathing in conjunction with worsening of pulmonary mechanics led to inefficient clearance of CO₂ and thus to acute hypercapnia (23).

The systemic hemodynamic effects of acute hypercapnia observed in the failure weaning group, i.e., increase in cardiac frequency, mean arterial pressure, and cardiac index, are in accordance with previous studies (24). The hypothesis of an increased sympathetic activity and so endogenous catecholamine release associated with weaning failure has been well documented (25).

Gastric Mucosal Acid-Base Status and Blood Flow

Gastric intramucosal acidosis has been proposed to be a sensitive index of gastrointestinal hypoperfusion (7). Elizalde and coworkers studied gastric mucosal perfusion by laser-Doppler flowmetry and reflectance spectrophotometry in 17 mechanically ventilated patients with stable hemodynamic status (26). Patients with intramucosal acidosis had a significant lower gastric mucosal blood flow than did nonacidotic patients, supporting the hypothesis that gastric mucosal hypoperfusion underlies the development of intramucosal acidosis in mechanically ventilated patients.

Gastric intramucosal pH has been proposed as a predictor of success or failure in weaning patients from mechanical ventilation (8, 9). Mohsenifar and coworkers measured the PCO₂ of gastric juice in 29 critically ill patients during a PS weaning trial. A gastric mucosal pH < 7.3 before the weaning attempt or a decrease by 0.09 or more during a 30-min weaning attempt was successful in predicting all weaning failures, with a better predictive value than any other conventional weaning parameters (8). This pH_i threshold value of 7.30 was confirmed by Bouachour and coworkers (9) during a 20-min spontaneous breathing trial in 26 patients with COPD. In addition, significant differences in pH_i and PCO₂im were observed between weaning success and weaning failure patients during mechanical ventilation before the weaning trial (9). Several elements may explain differences observed with our study. First, our weaning failure group is characterized by the appearance of acute respiratory acidosis during the 2-h weaning attempt (Table 2), which was not the case in the Mohsenifar and Bouachour studies. Moreover, the aim of our study was not to show the predictive value of gastric intramucosal pH during weaning but to observe changes in tonometric data and their relation with GMBF during weaning from mechanical ventilation, especially when weaning failure occurred. The weaning trial time used in these two studies was 30 min, which could be considered a short time for the appearance of acute hypercapnia in weanable patients with COPD. Second, tonometric data in the study of Mohsenifar were obtained directly by sampling the gastric juice; in this study, the gastric intramucosal PCO₂ values obtained in the failure weaning group were high and much greater than those reported in many other studies, even in groups with severe states and high mortality (27). These high values of gastric juice PCO₂ may be explained by reflux of bicarbonate-rich duodenal fluid into the stomach, which constitutes an important source of error in the measure of intramucosal PCO₂ (14).

In our study, patients who failed a weaning trial developed gastric intramucosal acidosis. However, in those patients who developed respiratory acidosis, a significant correlation was observed between arterial and intramucosal PCO₂, which is the main determinant of calculated pH_i (Figure 2). In an experimental model of acute hypercapnia by rebreathing method, Fiddian-Green and coworkers found that gastric PCO₂ equilibrated over a period of 1 h with arterial PCO₂ (15). Because the gastric intramucosal-to-arterial PCO₂ difference did not vary in the two groups, variations in gastric intramucosal pH are mainly due to arterial PCO₂ variations rather than gastric mucosal critical underperfusion (11).

Nevertheless, gastric mucosal blood flow variations do differ significantly according to weaning success or failure, and this may be due to differences in blood flow redistribution capabilities between these two groups. In the failure weaning group, a decrease in GMBF is observed during rapid shallow breathing and respiratory acidosis. Increased work and, thus, increased oxygen cost of breathing is well documented during weaning from mechanical ventilation (4). In the absence of variation in arterial oxygen delivery index, an increase in oxygen delivery to respiratory muscles requires a redistribution of organ blood flow. In a study of respiratory failure induced in dogs, Magder and coworkers found that respiratory muscle blood flow was increased in response to enhanced respiratory work while splanchnic and renal blood flow decreased (6).

However, beside this explanation, one may also hypothesize that changes in transdiaphragmatic pressure and especially in intrathoracic pressure regimen explain our findings (30). Rapid shallow breathing in patients with COPD who failed weaning is responsible for hypercapnia and increased intrinsic positive end-expiratory pressure (PEEP_i) (23). The presence of a residual positive expiratory pressure is known to impede splanchnic organ blood flow (28, 29). In an experimental study including 10 dogs without lung injury, Beyer and coworkers studied the effects of external PEEP on organ blood flow, with a radioisotope-labeling technique: total splanchnic blood flow was reduced in parallel with the fall in cardiac output, but the gastric perfusion declined out of proportion (29). On the other hand, the abdominal pressure and thus the surrounding pressure to the intraabdominal vasculature may influence the splanchnic circulation (30). Increased abdominal activity secondary to expiratory muscle recruitement has been demonstrated in resting patients with COPD (31) and during weaning failure (32). No specific measurement of gastric pressure was recorded in our study, but it is likely that this increased abdominal pressure during weaning failure should have regional circulatory effects on GMBF measurements. In the success weaning group, an increase in GMBF was observed during the PS trial while patients did not present any sign of respiratory failure. Volume-targeted assist-control ventilation increases right atrial pressure and, in a smaller proportion, abdominal pressure by compressing the abdominal contents (33). This could lead to a reduction of the driving pressure for venous blood flow from the abdomen and thus splanchnic circulation to the right side of the heart (33). When switching from volume-targeted assist-control ventilation, PS ventilation leads to lower pressure in the airway, a lower I:E ratio (34), and lower PEEP_i (35). Some could hypothesize that this decrease in the back pressure for venous return during a PS weaning trial may explain the rise in GMBF observed in the success weaning group.

In conclusion, during a 2-h pressure support weaning trial in patients with COPD, changes in gastric intramucosal pH do occur and differ according to weaning success or failure, but these changes appears to be linked mainly to systemic arterial PCO₂ variations. Nevertheless, gastric mucosal blood flow does vary during such a weaning trial according to the weaning issue. Patients who fail the weaning trial experiment experience a decrease in GMBF whereas GMBF increases in patients who succeed. This pattern may be due to a difference in blood flow redistribution to respiratory muscles because of a difference in energy requirements but may also be explained by a difference in transdiaphragmatic pressure patterns.