Effects of Inverse Ratio Ventilation versus  Positive End-expiratory Pressure on Gas Exchange and Gastric Intramucosal Pco²and pH under Constant Mean Airway Pressure in Acute Respiratory Distress Syndrome

The study was approved by the Institutional Review Board for Human Research of Chang Gung Memorial Hospital (Taoyuan, Taiwan). Informed written consent was obtained from the nearest relative, who had been given a full explanation of the study. A total of 17 ARDS patients with a lung injury score 19greater than 2.5 (age, 57 ± 21 y [mean ± SD]; range, 23–90 y; 9 men, 8 women) who were admitted to the medical intensive care unit were enrolled. Criteria for inclusion were as follows: (1) acute onset lung injury, (2) arterial oxygen tension/fraction of inspired oxygen (Pao²/Fio²) ≤ 200 mmHg, (3) bilateral infiltrates seen on the frontal chest radiograph, and (4) pulmonary wedge pressure ≤ 18 mmHg. 20Only patients whose hemodynamics were judged to be clinically stable were enrolled, although vasoactive drugs, such as dopamine and norepinephrine, were still needed in 14 of the 17 patients. All the patients were observed early in the course of ARDS. The duration from diagnosis of ARDS to the time of study was 2.5 ± 1 days.

Hemodynamics, including systolic arterial pressure, mean systemic arterial pressure, pulmonary arterial pressure, central venous pressure, and heart rate, during the time of blood gas sampling were recorded from an on-line HP Component Monitoring System (model 56S, M 1165A; Hewlett-Packard, Boblingen, Germany). The cardiac output computer (Vigilance Monitor; Baxter Edwards Critical Care, Irvine, CA) used in this study, along with the pulmonary artery catheter (IntelliCath, 7.5-French; Baxter Edwards Critical Care), could monitor cardiac output automatically and continuously using thermodilution principles and stochastic system identification techniques. The displayed value during blood gas sampling was recorded to represent the cardiac output at the time of that blood gas analysis.

Arterial and mixed venous blood gas samples were obtained simultaneously from the arterial cannula and pulmonary artery catheter at the end of each ventilatory setting, and kept anaerobically for analysis with a blood gas analyzer (Corning 178; Ciba Corning Diagnostic Corp., Medfield, MA) as soon as possible. Ḋo²and V̇o²were calculated as

and

respectively. The venous admixture (Q^S/Q^T) was calculated as

Cao²and Cvo²represent the oxygen content of the systemic artery and mixed venous blood, respectively, and were calculated as

Cc′o²represents the oxygen content of end-capillary blood and was derived from the same equation. Capillary oxygen partial pressure (Pc′o²) was assumed to be equal to the alveolar oxygen partial pressure (PAo²).

A gastric tonometer catheter (TRIP NGS catheter; Tonometrics, Inc. Worcester, MA) was inserted into the stomach via  the nasogastric route. Correct positioning of the catheters in the stomach was confirmed radiographically and by an auscultation of injected air over the epigastrium. The tonometer catheter was connected to an automated gas analyzer (Tonocap; Datex, Helsinki, Finland). The Tonocap monitor automatically fills the tonometer balloon with 5 ml room air, which is then kept in the catheter to allow carbon dioxide diffusion. After a preset equilibration time of 15 min (default), the system automatically aspirates the air sample from the catheter and measures partial pressure of carbon dioxide (Pco²) with an infrared sensor. The Tonocap monitor calculates the pHi from the arterial carbon dioxide tension (Paco²) and arterial pH (pHa) entered by the user and the latest Pico²value measured, using the following formula:

The air is then recycled into the catheter balloon to avoid carbon dioxide depletion in the catheter system. The entire sequence of filling, aspiration, Pico²measurement, and reinflation is fully automated. 21 

Airway pressure, flow, and esophageal pressure were continuously measured using a Bicore CP-100 pulmonary mechanics monitor (Bicore, Irvine, CA). The pressure transducer with pneumotachograph (VarFlex Flow Transducer; Bicore, Irvine, CA) was placed between the Y piece and the endotracheal tube. Peak and mean airway pressures were breath-by-breath and were continuously displayed by the Bicore CP-100 monitor. The proper position of the esophageal balloon (SmartCath Esophageal Balloon Catheter; Bicore, Irvine, CA) in the mid esophagus was confirmed radiographically and by cardiac oscillations observed on the esophageal pressure tracing. 22The Bicore CP-100 monitor automatically filled the balloon with 0.8 ml air and performed a leakage test. Tidal volume (V^T) was obtained by numerical integration of the inspiratory flow signal. Minute ventilation and respiratory rate were electronically analyzed and calculated from the flow signal. End-inspiratory plateau and end-expiratory plateau pressures were established by activating the inspiratory or expiratory hold on the Servo 300 ventilator (Servo 300 ventilator; Siemens-Elema, Solna, Sweden) for 5 s. These occlusion maneuvers were repeated 3–5 times, leaving five regular mechanical ventilatory cycles in between. Auto-PEEP was defined as the difference between total PEEP and applied PEEP. All the digital parameters derived from the Bicore CP-100 were recorded continuously, breath-by-breath, and were stored in a personal computer in the spreadsheet form for 5 min at the end of each ventilatory setting. To eliminate probable spontaneous variation, the mean values of these breaths were used to represent the ventilatory status of that period.

All patients were sedated with a continuous infusion of midazolam (F. Hoffmann-La Roche Ltd., Basle, Switzerland), propofol (ICI Pharmaceuticals, Cheshire, England), or both and were paralyzed with atracurium (F H Faulding & Co. Ltd., Victoria, Australia) to avoid muscular activity during the study. They were ventilated at a rate of 20 breaths/min with a Servo 300 ventilator (Siemens-Elema). The inspiratory pressure was set to deliver a V^Tof 6–8 ml/kg. To determine the optimal level of MAP, we adjusted the level of applied PEEP from the original setting by an increment of 2–3 cm H²O each time. According to the recommendation of Patel and Singer, 23Ḋo²was measured 15 min after each increase of applied PEEP. The level of PEEP was increased gradually until Ḋo²reached a plateau or even decreased. The MAP at the PEEP level with the highest Ḋo²was regarded as the optimal MAP. The optimal MAP and V^Twere held constant by adjusting the levels of peak inspiratory pressure and applied PEEP after each change of I:E ratio. Fio²was kept the same as its previous use before entry into the study and ranged from 40 to 100%.

The study comprised five ventilatory settings of 60 min each. The initial ventilatory mode was the PCV with a conventional I:E of 1:2. The ventilatory setting was then shifted to PC-IRV 1:1. We anticipated that the dynamic hyperinflation and auto-PEEP might become greater and that the hemodynamic compromise would become more severe after the I:E ratio became greater than 1:1. Therefore, after the low-grade inverse of I:E 1:1, patients were randomized to receive PC-IRV 2:1, then 4:1 or 4:1, and then 2:1, alternately. To eliminate the probable sustained IRV effects from the previous mode, the baseline setting PCV 1:2 was repeated between settings of PC-IRV 2:1 and 4:1. The sequence of the ventilatory settings, as a result, was either sequence A: baseline PCV 1:2 (1) → PC-IRV 1:1 → PC-IRV 2:1 → baseline PCV 1:2 (2) → PC-IRV 4:1; or sequence B: baseline PCV 1:2 (1) → PC-IRV 1:1 → PC-IRV 4:1 → baseline PCV 1:2 (2) → PC-IRV 2:1. In each ventilatory mode, measurements of hemodynamics, pulmonary mechanics, arterial and mixed venous blood gas analysis, Pico², and pHi were performed after a 1-h period of stabilization. The rates of intravenous fluid and vasopressor infusion were kept fixed during the study periods. Endotracheal suctioning was performed only after the recording of the investigated parameters.

The results obtained are expressed as mean ± SD. A P  value less than 0.05 was considered significant. The 95% confidence interval 24of the difference between data from each of the three PC-IRV settings and its previous baseline PCV 1:2 (Δ1:1, Δ2:1, Δ4:1) is calculated to compare the effects of PC-IRV with PCV 1:2. The Δ1:1 represents the difference between PC-IRV 1:1 and baseline PCV 1:2 (1). The Δ2:1 represents the difference between PC-IRV 2:1 with baseline PCV 1:2 (1) for patients with sequence A and the difference between PC-IRV 2:1 with baseline PCV 1:2 (2) for patients with sequence B. On the contrary, Δ4:1 represents the difference between PC-IRV 4:1 with baseline PCV 1:2 (2) for patients with sequence A and the difference between PC-IRV 4:1 with baseline PCV 1:2 (1) for patients with sequence B. The 95% confidence interval represents the range between mean ± (t¹-α^/2× SD/ⁿ>), where t¹-α^/2is the appropriate value from the t  distribution with n − 1 degrees of freedom associated with a confidence of 100 × (1-α)%. A zero difference outside the 95% confidence interval indicated that a statistically significant difference existed between the setting of PC-IRV with its previous baseline PCV 1:2 at the 5% level.

Table 1summarizes the main characteristics of the patients. The severity of ARDS expressed as lung injury score was 3.2 ± 0.4 (range, 2.5–3.75). The mean APACHE III score at the time of diagnosing ARDS was 74 ± 26 (range, 42–120). Eight cases were studied with sequence A, and nine cases were studied with sequence B. The initial applied PEEP level set to achieve the optimal MAP was 14.5 ± 2.8 cm H²O (range, 10–20 cm H²O). In case 6, the exact values of Pico²measured were too high in all settings of PC-IRV, higher than the upper limit (112 mmHg) displayed by the Tonocap monitor. Therefore, only 16 cases were studied in the comparisons of pHi and Pico². All patients completed the study uneventfully.

With the MAP kept constant, PC-IRV did not compromise the hemodynamics, regardless of the degree of IRV (table 2). There were no significant changes in cardiac output, heart rate, systolic and mean arterial pressures, central venous pressure, and pulmonary capillary wedge pressure throughout the study. Systemic and pulmonary vascular resistance were stable, as well. The systolic and mean pulmonary arterial pressures during PC-IRV 4:1 were greater than those during baseline PCV 1:2, but the differences were small.

Arterial and mixed venous blood gases analysis, global oxygenation indices of Ḋo²and V̇o², and regional oxygenation index of Pico²and pHi during ventilation with all the various I:E ratios are listed in table 3. PC-IRV 2:1 and 4:1 significantly worsened both Pao²and mixed venous oxygen tension (Pvo²) as compared with its previous baseline PCV 1:2. Q^S/Q^Tduring PC-IRV 4:1 was slightly higher than during PCV 1:2. Ḋo²and V̇o²remained stationary after shifting the mode to PC-IRV. The high-grade IRV (2:1 and 4:1) resulted in a significant increase in pHa and decreases in Paco², Pvco², and bicarbonate (HCO³⁻). Despite the changes in pHa, Paco², and HCO³⁻, Pico²and pHi remained unchanged across all five settings. The gastric intramucosal–arterial Pco²gradient (Pico²–Paco²) during PC-IRV 1:1 is greater than the Pico²–Paco²gradient during PCV 1:2.

In keeping with the study design, MAP, V^T, and minute ventilation were kept similar. PC-IRV resulted in a progressively greater reduction in peak inspiratory pressure than did the previous baseline PCV 1:2. Auto-PEEP augmented gradually with the extension of inspiratory time. To maintain MAP at a constant level, the applied PEEP and total PEEP levels needed to be reduced accordingly after shifting the ventilator to PC-IRV.

The major findings of this study are that at constant mean airway pressure, PC-IRV did not improve arterial oxygenation, global oxygenation indices (Ḋo²and V̇o²), and gastric intramucosal Pco²and pH more than did PCV 1:2 with PEEP.

Mean airway pressure, as a clinically measurable reflection of mean alveolar pressure, fundamentally reflects oxygen exchange and cardiovascular performance during conditions of passive inflation. 11,12,14Excessively high MAP, although improving arterial oxygenation, may decrease cardiac output and jeopardize Ḋo². Consequently, MAP should be increased only sufficiently to accomplish the unequivocal clinical objectives for gas exchange and maximal Ḋo². Recently, in several studies in which IRV was compared with conventional ratio ventilation with similar total PEEP (applied PEEP plus auto-PEEP), 2,6,8,9,25the increase of MAP after implementation of IRV resulted in a decrease of cardiac output. This may then contribute to a further decrease of Ḋo²and implies that MAP might have been increased to a deleterious level. Therefore, MAP should be kept constant when settling the clinical utility of IRV over applied PEEP.

With MAP kept constant, previous reports comparing IRV with VCV demonstrated no change in cardiac output. 25–27Similarly, our data showed that cardiac output, arterial blood pressure, and vascular resistance were stationary throughout the study. PC-IRV 4:1 only resulted in small changes in systolic and mean pulmonary artery pressures (table 2). Using applied PEEP or IRV to reach the same MAP exerted the same influences on cardiac output and hemodynamics.

The effects of different I:E ratios on arterial oxygenation might result from the contrasting effects between applied PEEP and auto-PEEP. The amount of applied PEEP adds directly to the level of MAP, whereas during IRV, MAP increases because mean inspiratory pressure increases and because auto-PEEP is created. 11,12To maintain a constant MAP, the level of applied PEEP after progressive lengthening of inspiratory time must be reduced. On the contrary, auto-PEEP augmented gradually. Although applied PEEP is homogeneously distributed, Kacmarek et al.  28found that a greater maldistribution of local lung unit end-expiratory pressure was established with the auto-PEEP in a four-unit lung model. Because parenchymal involvement in ARDS is heterogeneous, the distribution of auto-PEEP induced by PC-IRV is likely to be uneven because of time constant inequalities between lung units. Therefore, the slowest lung units could still be collapsed while a substantial level of auto-PEEP is measured by the end-expiratory occlusion method. However, the local PEEP for the fast lung units might be inadequate because of the decrease of applied PEEP. Steward et al.  29demonstrated that applied PEEP is more effective in improving Pao²than IRV per increment of MAP (ΔPao²/ΔMAP 6.08 vs.  1.90). Pesenti et al.  25disclosed that Pao², with MAP kept constant by extending the inspiratory time, was much decreased when an applied PEEP lower than inflection-point pressure (P^flex) was compared with an applied PEEP higher than P^flex. Brandolese et al.  30also discovered a more significant improvement in arterial oxygenation with applied PEEP than with auto-PEEP, which they attributed to the less homogeneous distribution of auto-PEEP between lung units with different time constants.

When comparing the cardiorespiratory effects of PCV with PC-IRV in an acute lung injury animal model, Mang et al.  27adjusted the pressure control level and applied PEEP to maintain a constant V^Tand MAP. Their data showed that Pao², albeit not statistically significant, was lower with PC-IRV 4:1 than with PCV 1:2 (172 ± 69 vs.  201 ± 78 mmHg). Performed in ARDS patients, our study corroborated the findings of Mang et al.  27Pao²and Pvo²both decreased significantly when the inspiratory time was prolonged to PC-IRV 2:1 and 4:1 (table 3). The decreased applied PEEP levels (table 4), with the accompanying higher Q^S/Q^T(table 3), probably accounted for the deterioration of Pao²with PC-IRV. In accordance with the aforementioned study, under similar MAP, replacing the applied PEEP with auto-PEEP from IRV was detrimental to both arterial and mixed venous oxygenation in ARDS patients.

Extending the inspiratory time enhances the clearance of carbon dioxide. PC-IRV has been found to improve Paco²more than PCV 2,6or volume-cycled ventilation with PEEP do. 8This effect is usually attributed to a decrease in physiologic dead space and an improved mixing of alveolar and bronchial gases. Similarly, PC-IRV 2:1 and 4:1 significantly decreased Paco²more than PCV 1:2 did.

Gastric Pico²can increase as a result of respiratory acidosis, i.e. , stagnation of carbon dioxide with decreased washout from low mucosal blood flow, or as a result of metabolic acidosis, i.e. , buffering of the hydrogen ion (H⁺) from anaerobic metabolism by HCO³⁻. 16Recently, Elizalde et al.  31demonstrated that impairment of gastric mucosal blood flow underlies the development of intramucosal acidosis in mechanically ventilated patients and contended that the measurement of gastric pHi constituted a method to assess the adequacy of perfusion of the gastrointestinal tract in critically ill patients. Knichwitz et al.  32proved that compromised mesenteric blood flow causes significant metabolic and histologic changes in the gastrointestinal tract and argued that the only parameter of importance is the intraluminal measurement of intramucosal Pco²that can reflect isolated mesenteric change. Regardless of the origin of increased Pico², intramucosal acidosis has been clinically associated with a grave prognosis for intensive care unit patients. Pico²and pHi have been used as an index to judge the effectiveness of therapeutic interventions.

Inadequate splanchnic perfusion may be caused by decreased total cardiac output, inappropriate distribution of flow between organs, and mismatching of oxygen supply to demand within organs. PEEP has been proved not only to reduce cardiac output, 33but also to decrease mesenteric and portal blood flow 34–35and to redistribute the cardiac output away from the splanchnic circulation. 36Therefore, although beneficial in improving arterial oxygenation, the application of PEEP may contribute to mesenteric ischemia and gastrointestinal failure caused by alteration in regional blood flow. The interrelation between IRV and splanchnic blood flow has not been clearly documented. Under constant MAP and V^T, IRV reduced the peak inspiratory pressure and total PEEP needed. However, in this study, we did not detect any significant fluctuation in either Pico²or pHi during PC-IRV as compared with baseline PCV 1:2 (table 3). Because changes in systemic Pco²produced by systemic respiratory acidosis or alkalosis directly and rapidly affected Pico², the difference between Pico²and Paco²(Pico²–Paco²) has been suggested to be more informative and to be a more reliable index of gastric mucosal oxygenation than is pHi or Pico²alone because of the correction for abnormalities in systemic arterial Pco². 37 Table 3shows that Pico²–Paco²in PC-IRV 1:1 was significantly greater than in baseline PCV 1:2 with high PEEP. Instead of improvement, PC-IRV did not mitigate gastric intramucosal acidosis or improve the mucosal hypoperfusion.

In summary, our data do not support the application of IRV as a substitute for applied PEEP to achieve a similar MAP in ARDS patients. If we set MAP to be optimal, i.e. , with maximal Ḋo², by titrating the levels of applied PEEP, PC-IRV does not contribute to better gas exchange and gastric intramucosal Pco²and pH than does PCV 1:2 with applied PEEP, regardless of the I:E ratios.