Pressure-release Tracheal Gas Insufflation Reduces ... - ATS Journals

Pressure-release Tracheal Gas Insufflation Reduces Airway Pressures in Lung-injured Sheep Maintaining Eucapnia MAX KIRMSE, YUJI FUJINO, JONATHAN HROMI, HARALD MANG, DEAN HESS, and ROBERT M. KACMAREK Respiratory Care Department Laboratory and the Department of Anesthesia and Critical Care, Massachusetts General Hospital and Harvard Medical School, Boston, Massachusetts

Although tracheal gas insufflation (TGI) has proved to be a useful adjunct to mechanical ventilation, end-inspiratory as well as end-expiratory pressures may increase. We investigated the ability of continuous-flow TGI to maintain eucapnia while reducing airway pressure (Paw) and tidal volume (VT). Seven sheep (36 6 2 kg) were ventilated using the Dräger Evita 4 in the pressure control plus mode where flow is released via the expiratory valve to maintain constant inspiratory pressure. To avoid TGI-generated positive end-expiratory pressure (PEEP), a prototype reverse flow TGI tube was used. Two TGI flows (5 and 10 L/min) were investigated pre- and postsaline lavage–induced lung injury. Inspiratory pressures and VT were significantly reduced as TGI flow increased. At 10 L/min TGI flow the carinal pressures (Pcar) and VT were reduced pre- and postinjury by 15% and 20%, and by 28% and 34%, respectively. Tidal volume to dead space ratio (VD/VT) decreased preinjury from 0.49 6 0.1 to 0.18 6 0.2 and postinjury from 0.62 6 0.1 to 0.33 6 0.1 at a TGI flow of 10 L/min. The combination of the reverse flow TGI tube and a ventilator with an inspiratory pressure relief mechanism kept set end-inspiratory and end-expiratory pressures constant. This TGI system effectively reduced set Paw and VT while maintaining eucapnia. Kirmse M, Fujino Y, Hromi J, Mang H, Hess D, Kacmarek RM. Pressure-release tracheal gas insufflation reduces airway pressures in lung-injured sheep maintaining eucapnia. AM J RESPIR CRIT CARE MED 1999;160:1462–1467.

Ventilatory strategies for patients with acute respiratory distress syndrome (ARDS) should provide gas exchange without risk of ventilator-induced lung injury (1, 2). Current recommendations limit the inspiratory pressure to decrease the risk of alveolar overdistention (3). This can be accomplished by decreasing tidal volume (VT) at the cost of an increased PaCO2 (1, 2) or by improving alveolar ventilation and CO2 removal (4, 5). Tracheal gas insufflation (TGI) has been proposed as an adjunct to mechanical ventilation to reduce the dead space (VD) of the patient/ventilator system, allowing smaller VT (4, 5). During TGI, gas is insufflated near the carina via a catheter (4, 5) or through specially designed endotracheal tubes (6). This secondary gas flow flushes the proximal VD of CO2. This allows reduced VT and airway pressures (Paw) while limiting CO2. TGI has primarily been used with controlled hypoventilation—which already limits VT—but leads to an increase in (Received in original form January 12, 1999 and in revised form April 15, 1999)

arterial CO2 (4–8). However, TGI may also be useful in patients who cannot tolerate elevated CO2 levels (e.g., head trauma, cardiac dysfunction). Technical and safety issues make TGI potentially dangerous. During continuous-flow TGI (C-TGI) excessive inspiratory pressure may develop (9), which can be eliminated by injecting the TGI flow only during the expiratory phase (Ex-TGI) (3, 4, 10, 11). TGI may also increase the end-expiratory pressure when the TGI flow is directed toward the carina (12, 13). Although this increase in positive end-expiratory pressure (PEEP) can be corrected by decreasing the level of applied PEEP, monitoring total PEEP during TGI is difficult (5). In this study we evaluated the ability of C-TGI using a prototype reverse-flow TGI endotracheal tube and the Dräger Evita 4 ventilator (Dräger Inc., Chantilly, VA) to maintain eucapnia while reducing Paw and VT in both noninjured and lung-injured sheep. The Evita 4 maintains targeted pressure constant via its active exhalation valve during both inspiration and expiration.

Supported in part by Sherwood, Davis & Geck. Dr. Mang has a financial interest in Sherwood, Davis & Geck. Dr. Kirmse was supported in part by a grant from the Deutsche Forschungsgemeinschaft (KI 582/1-1). Dr. Fujino was supported in part by the Japanese government. Correspondence and request for reprints should be addressed to Robert M. Kacmarek, Ph.D., Respiratory Care–Ellison 401, Massachusetts General Hospital, Boston, MA 02114. Am J Respir Crit Care Med Vol 160. pp 1462–1467, 1999 Internet address: www.atsjournals.org

METHODS The following protocol was approved by the Subcommittee on Research Animal Care of the Massachusetts General Hospital.

Preparation Seven fasted Hampshire sheep (36 6 2 kg body weight) were anesthetized with halothane, and orally intubated with an orogastric tube inserted. After cannulating the left jugular vein and a loading dose of

1463

Kirmse, Fujino, Hromi, et al.: Pressure-release Tracheal Gas Insufflation fentanyl (0.3 mg) and diazepam (10 mg), anesthesia was maintained using fentanyl (3 mg/kg/h), sodium pentobarbital (5 mg/kg/h), and pancuronium bromide (2 mg/h). A tracheotomy was performed and the airway was cannulated with a prototype double-lumen TGI tube (6) with its tip 2 to 3 cm from the carina. Alongside the TGI tube a 0.9mm-diameter catheter was placed for carinal pressure (Pcar) measurements. The tip of the catheter extended 1 cm beyond the tip of the TGI tube (verified bronchoscopically). Using the Evita 4 ventilator (Dräger Inc.) the sheep were ventilated in the pressure control plus mode (PCV1), fraction of inspired oxygen (FIO2) 0.5, rate 20/min, inspiratory to expiratory ratio (I:E) 1:2, and PEEP 10 cm H2O. Inspiratory pressures were set to provide normocapnia (PaCO2 approximately 40 mm Hg). The right femoral artery was cannulated and a pulmonary artery catheter (Edwards Swan Ganz, 7.0 French; Baxter Healthcare Corp., Irvine, CA) was placed via the right jugular vein. Lactated Ringer’s solution was administered to maintain central venous pressure (Pcv) between 6 and 12 mm Hg. Core temperature was maintained at 398 C using an electric heater and blankets.

TGI System The TGI tube is a prototype double-lumen endotracheal tube (Sherwood, St. Louis, MO) with a 7.8-mm interior diameter (i.d.) main lumen, and a second TGI insufflation lumen of 3.5 mm i.d. (Figure 1). The TGI gas flow is directed cephalad into the main lumen of the tube by a nozzle at the tip of the tube (reverse-flow TGI) (6). TGI gas flow was continuously delivered (C-TGI) via an oxygen blender (Bird/3M, Palm Springs, CA) set to the FIO2 delivered by the ventilator (Figure 1). The TGI flows (5 and 10 L/min) were verified with a precision calibration flowmeter. To measure total PEEP, a solenoid valve (model A3314-S8; Precision Dynamic Inc., New Britain, CT) was placed between the flowmeter and the TGI tube. The valve was computer synchronized with the ventilator and diverted the TGI flow at end-expiration to atmosphere (6) when the self-controlled PEEP (auto-PEEP) function was engaged on the Evita 4. Although the TGI flow was continuous, the Evita 4 prevented pressure from exceeding the set inspiratory pressure (Figure 2).

Experimental Protocol Two TGI flows (5 and 10 L/min) were investigated before and after lung injury. Before and after each randomly applied TGI flow, a con-

trol measurement without TGI was performed. Pressure control levels were set to provide normocapnia (PaCO2 approximately 40 mm Hg), and were decreased from control levels when TGI was applied. The PEEP set on the ventilator was adjusted to keep a constant total PEEP of 10 cm H2O. Each experimental stage was maintained for 30 min, after which blood gases 5 min apart were drawn to verify steady state, and measurements were obtained.

Lung Injury After five nonlung-injured evaluations (three control, two TGI), FIO2 was set to 1.0, and bilateral lung lavage was performed with isotonic saline (30 ml/kg body weight) warmed to 398 C (14) and repeated every 15 min until the PaO2/FIO2 was , 100 mm Hg. After lung lavage, the sheep were allowed to stabilize for 90 min while the pressure control level was adjusted to a new baseline. After control measurements the protocol was repeated for the lung-injured sheep.

Measurements At each step (control, TGI 5, TGI 10) before, and after lung injury hemodynamics, gas exchange, and pulmonary mechanics were measured. Arterial pressure (Pa), pulmonary artery pressure (Ppa), and Pcv were monitored using pressure transducers (049924-507A, Argon; Maxxim Medical, Athens, TX) calibrated at 50 mm Hg with the zero level at midthorax in the supine position. Pulmonary artery occlusion pressure .(Ppao) and Pcv were measured at end-expiration. Cardiac output (Q) was measured by thermodilution (9520A; American Edwards Laboratories, Irvine, CA) as the average of three sequential measurements obtained by injecting 5 ml of 08 C lactated Ringer’s solution. At the end of each evaluation period, arterial and mixed venous blood gases were assessed (model 238, Ciba Corning Diagnostics Corp., Norwood, MA; and model 282, Instrumentation Laboratory, Lexington, MA). Paw and Pcar were monitored with pressure transducers (model 45-32-871 6 100; Validyne, Northridge, CA) calibrated with a water manometer. Inspiratory and expiratory flows were measured separately by two pneumotachometers (3700A; Hans-Rudolph Inc., Kansas City, MO) in each limb of the ventilator circuit. The two pneumotachometers were calibrated in series. One flow signal was generated by adding both flow signals, correcting for compressible volume. Inspiratory VT were obtained by integration of this flow signal and confirmed with a 500-ml calibration syringe. The effective inspiratory

Figure 1. The experimental setup. The FIO2 of the TGI gas flow was adjusted with a blender and then continuously delivered to the TGI endotracheal tube. The solenoid valve diverted TGI flow to atmosphere when total PEEP was measured by the ventilator. The TGI flow was injected into the main lumen of the · · TGI tube. Pneumotachometers measured inspiratory (VIn) and expiratory flows (VEx). Pressures were measured at the carina (Pcar) and at the Y-piece of the ventilator circuit (Paw).

1464

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE

VOL 160

1999

TABLE 2 AIRWAY MECHANICS DURING CONTROL AND TGI 5 AND 10 L/MIN, BEFORE (FIO2 0.5) AND AFTER LUNG INJURY (FIO2 1.0)*

Preinjury Paw, cm H2O Pcar, cm H2O PEEP, cm H2O VT, ml VD/VT Injury Paw, cm H2O Pcar, cm H2O PEEP, cm H2O VT, ml VD/VT

Control

TGI 5

TGI 10

23.0 6 1.6 22.6 6 1.6 10.0 6 0.2 272 6 46 0.49 6 0.10

21.0 6 1.9† 20.4 6 1.9† 9.9 6 0.2 225 6 45 0.36 6 0.15†

20.7 6 2.0† 19.2 6 1.7†‡ 10.0 6 0.6 196 6 34†‡ 0.18 6 0.19†‡

31.9 6 4.9§ 31.4 6 4.8§ 10.1 6 0.3 322 6 33 0.62 6 0.06

27.8 6 4.8† 27.1 6 4.7† 10.1 6 0.2 255 6 45† 0.51 6 0.07†

26.7 6 5.4† 25.2 6 5.2†‡ 10.1 6 0.6 212 6 46†‡ 0.33 6 0.12†‡

* Values are expressed as mean 6 SD. † Versus control, ‡ versus TGI 5, p , 0.05; one-way ANOVA repeated measures and Tukey’s post hoc. § Preinjury versus injury, p , 0.05; t test for dependent samples.

Figure 2. Tracings of flow and pressures, depicting the release of additional flow during inspiration (lung injury group, TGI flow 10 L/min). During inspiration the expiratory valve keeps the set Paw constant by releasing flow (TEXinsp). Lung deflation starts after the set inspiratory time (TI) and at the end of the expiratory time (TEX), · expiratory flow is constant and equals TGI flow ( VTGI). The difference between Paw (solid line) and Pcar (dashed line) is caused by the design of the TGI tube. Whereas Paw is increased above Pcar during inspiration, when the two gas flows from the ventilator and the TGI system are in counterdirection and because of resistance of the tube, the entrainment of the TGI tube during expiration keep Pcar below Paw.

tidal volume (VINeff) from the ventilator (VVENT) and the TGI system · was then calculated: VINeff 5 VVENT 1 VTGI 3 TI 2 VEXinsp where · VTGI is the TGI flow in L/s measured at end-expiration, TI is the inspiratory time in seconds, and VEXinsp is the total volume released via the exhalation valve during inspiration. To calculate dead space/tidal volume ratio (VD/VT) expired gas was collected for 3 min. The mean expired CO2 concentration (FECO2 )

in this bag was measured using a capnometer (model 2200; Traverse Medical Monitors, Saline, MI) and was corrected for the CO2 free TGI gas: FECO2 5 measured FECO2 3 (VEX 1 VEXinsp)/VEXeff, where VEX is the expiratory VT and effective exhaled VT (VEXeff) 5 VEX 2 · VTGI 3 TE. VD/VT was calculated according to the Enghoff’s modification of the Bohr equation (15) and CO2 production (VCO2) was cal· culated during baseline measurements as VCO2 5 FECO2 3 VE, where · minute ventilation ( VE) 5 VEXeff 3 respiratory rate. At each setting, total PEEP was measured with the end-expiratory hold maneuver of the Evita 4 and a computer controlling the solenoid valve to eliminate TGI flow. Compliance (CVent) was calculated for all baseline measurements using the end-inspiratory pressure at zero inspiratory flow (PPlateau): CVENT 5 VINeff/(PPlateau 2 PEEPTotal). All signals were amplified (model 8805C; Hewlett Packard, Waltham, MA) and recorded at 100 Hz using an A/D-conversion system (WINDAQ/200 V1.36; Dataq Instruments, Akron, PA).

Statistical Analysis All data are expressed as mean 6 standard deviation (SD) unless otherwise stated. For pulmonary mechanics measurements, three breaths were analyzed and averaged. For comparisons between the groups and the two TGI flows, a t test was applied. Within groups, one-way analysis of variance (ANOVA) for repeated measures was performed. Because no differences among the three control stages for each of the tested parameters in each group (preinjury/injury) were detected, their means were used for subsequent analysis. A one-way ANOVA for repeated measures compared control, TGI-5, and TGI10. When statistical significance was reached a post hoc analysis (Tukey honest significant difference [HSD] test) was performed. A statistics software package (STATISTICA 5.1; StatSoft Inc., Tulsa, OK) was used, considering a p value , 0.05 as significant.

TABLE 1 PaCO2, PH, AND PaO2/FIO2 RATIO DURING CONTROL AND TGI 5 AND 10 L/min, BEFORE (FIO2 0.5) AND AFTER LUNG INJURY (FIO2 1.0)*

Preinjury PaO2/FIO2, mm Hg PaCO2, mm Hg PH Injury PaO2/FIO2, mm Hg PaCO2, mm Hg PH

Control

TGI 5

TGI 10

410.57 6 45.56 37.71 6 2.09 7.42 6 0.05

403.43 6 69.7 37.57 6 2.57 7.42 6 0.05

394.29 6 67.25 37.29 6 1.89 7.41 6 0.06

75.33 6 21.0† 38.29 6 0.73 7.38 6 0.05

76.57 6 26.78 38.29 6 1.50 7.38 6 0.05

* Values are expressed as mean 6 SD. † Preinjury versus injury, p , 0.05; t test for dependent samples.

78.29 6 23.8 38.30 6 1.25 7.37 6 0.05

RESULTS Control Measurements

For all variables the three control settings in each group were not significantly different. As a result, they were averaged and compared with the TGI 5 and TGI 10 settings. Preinjury and Injury

Blood gas data, VD/VT, and Paw and Pcar preinjury and postinjury are summarized in Tables 1 and 2. With injury, PaO2 decreased and VD/VT, Paw, and Pcar increased. Hemodynamic data were constant throughout the study (Table 3). Lung injury reduced CVENT from 21.6 6 2.0 ml/cm H2O to 15.1 6 2.3 cm H2O (p , 0.001).

1465

Kirmse, Fujino, Hromi, et al.: Pressure-release Tracheal Gas Insufflation TABLE 3 HEMODYNAMICS DURING CONTROL AND TGI 5 AND 10 L/min, BEFORE (FIO2 0.5) AND AFTER LUNG INJURY (FIO2 1.0)*

Preinjury Pa , mm Hg Ppa, mm Hg Ppao, mm Hg Pcv, mm Hg · Q , L/min Injury Pa , mm Hg Ppa, mm Hg Ppao, mm Hg Pcv, mm Hg · Q , L/min

Control

TGI 5

TGI 10

111 6 9.6 19.4 6 2.5 7.6 6 1.0 7.8 6 2.9 3.8 6 0.8

114 6 11.5 19.0 6 3.3 7.6 6 1.7 8.3 6 3.9 3.8 6 0.8

109 6 9.0 19.4 6 3.6 7.1 6 0.9 7.3 6 2.5 3.7 6 0.8

102 6 9.0 23.0 6 2.5 9.1 6 2.3 8.3 6 3.9 4.1 6 1.1

101 6 12.1 22.7 6 3.4 9.4 6 2.0 8.8 6 3.6 4.0 6 1.3

100 6 13.7 22.6 6 2.7 9.0 6 2.4 8.0 6 3.6 3.9 6 1.3

Definition of abbreviations: Pa 5 mean arterial blood pressure; Ppa 5 mean pulmonary artery pressure; Ppao 5 pulmonary artery occlusion pressure; Pcv 5 central venous · pressure; Q 5 cardiac output. * Values are expressed as mean 6 SD.

Effect of C-TGI

The pressures (Paw and Pcar) required to maintain eucapnia decreased significantly from baseline when TGI was started (Table 2 and Figure 3, p , 0.0005). Because of the direction of gas flow, the increase in TGI flow from 5 to 10 L/min resulted in a significant decrease in Pcar (p , 0.05). At all TGI flows, gas was released by the expiratory valve during the inspiratory phase. The reduced PCV1 pressure settings resulted in a significant decrease in VT and VD/VT with increasing TGI flows (Table 2, ANOVA, p , 0.05). The TGI flow marginally decreased total PEEP (maximal, 1 cm H2O). Compared with control settings in each group, all variables for gas exchange remained unchanged during both TGI flows regardless of the presence or absence of injury (Table 1). TGI—Flush Volume

The effective flush volumes during both TGI flows are illustrated in Figure 4. Because of the inspiratory release of gas, total flush volume was significantly greater during inspiration than during expiration (p , 0.05).

DISCUSSION The major findings of this study are as follows: first, continuous-flow TGI during PCV1 using a ventilator with an active

Figure 3. Carinal plateau pressures during control and TGI 5 and TGI 10 before (open squares) and after lung injury (closed squares) (mean 6 SEM, *versus control, #versus TGI 5, p , 0.05).

Figure 4. Comparison of the effective flush volumes for TGI 5 and TGI 10 during preinjury and injury. The open bars show the flush volumes measured during the expiratory phase only. The black bars show the total flush volumes including the gas released during inspiration. In all cases the inspiratory release of gas increased the flush volume significantly (p , 0.05). However, during TGI 5 little additional volume (approximately 23 ml, 14%) was released, whereas during TGI 10 flush volume increased by about 25% (approximately 80 ml). Note that the flush volumes for preinjury and injury are almost identical (not significant) (mean 6 SEM, *versus expiratory flush volume, p , 0.05).

exhalation valve allowed reduction in inspiratory pressures and VT in the normal and injured lung while maintaining normocarbia; second, the ability of TGI to reduce Paw was not affected by lung injury. Effects of TGI on End-inspiratory Pressures

Most studies have focused on the ability of TGI to reduce PaCO2, not on the reduction of Paw and VT at a constant PaCO2 (9–13). However, many patients do not tolerate even mild increases in PaCO2 (16). In our study, both pre- and postinjury Pcar were reduced at a TGI flow of 10 L/min by 15% and 20% and VT by 28% and 34%, respectively. The decrease in Pcar and VT was affected by the TGI flow (5 L/min , 10 L/min), although the efficiency of pressure and volume reduction was not diminished after lung injury when alveolar VD increased. Because catheter flush volume was similar pre- and postinjury (Figure 4), the efficacy of TGI was mainly determined by the slope of the respiratory system pressure–volume relationship. As compliance decreased similar improvements in ventilation resulted in larger pressure and VT changes postinjury. The previously documented decrease in efficiency of TGI, when alveolar VD increased (7, 20), was offset by the deterioration in lung function. The ability of TGI to allow pressure to be reduced may be the most important feature of this technique, especially in the setting of overdistension. The interdependence of the efficacy of TGI and lung function renders comparisons with other studies that kept PaCO2 constant (18–20) difficult. Nakos and coworkers (19) reported a 20% reduction in airway pressure and a 25% reduction in VT with 6 L/min continuous-flow volume-adjusted TGI in COPD patients with acute respiratory failure. However, their patients had highly compliant lung/thoracic systems (53.53 6 10.71 ml/ cm H2O) and PaO2/FIO2 ratios exceeding 160 mm Hg. In addition, they also used volume ventilation without a method of maintaining peak airway (carinal) pressure constant. Nahum and coworkers (20) also reported a significant reduction in Paw (13%) and VT (34%) using volume-adjusted C-TGI at 15 L/min. However, they investigated lung-injured dogs that were hypoventilated. Our results compare well with

1466

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE

these groups (19, 20) although our animals were eucapnic. It can also be speculated that the performance of our TGI system was enhanced by pressure and volume release during inspiration. Although the released TGI gas during inspiration did not contribute to the flush volume—since at this point no CO2 was in the proximal airway—turbulence generated at the tip of the endotracheal tube may have increased CO2 clearance distal to the tip of the tube. TGI Delivery System

Several aspects of TGI delivery are problematic, making routine use difficult and potentially dangerous. Because TGI flow is from a separate gas source, increased inspiratory pressures with volume or pressure ventilation are likely when TGI is delivered continuously without an exhalation valve active during inspiration. TGI delivery during the expiratory phase (10, 13, 17) or volume-adjusted TGI (10) are proposed solutions to this problem, but require additional technology or gas delivery adjustments every time ventilator settings are changed. An inspiratory pressure relief system in combination with pressure control ventilation (21, 22) may be a safe and effective method of delivering continuous-flow TGI, because even inadvertent increases in TGI flow will not result in increased airway pressures. The system we have described seems ideal because it utilizes the capabilities of a commercially available ventilator. The Dräger Evita 4 is one of the few ventilators that has an exhalation valve active during inspiration in the PCV1 mode. The difference between PCV1 and PCV on other adult ventilators during controlled ventilation is an active exhalation valve during inspiration. The only other ventilator that has an exhalation valve active during inspiration in the PCV mode is the Nellcor Puritan Bennett 840 ventilator (Carlsbad, CA). In all other ventilators, the potential for Paw to increase during PCV and C-TGI exists. In these ventilators increases in pressure are released without cycling to the expiratory phase. In addition, all of the alarms and monitors of the ventilator are active. The single safeguard not available is Pcar monitoring and TGI flow shut off if endotracheal tube occlusion should occur. Direction of TGI Insufflation

Many TGI catheters (6, 23) and tube designs (24) have been evaluated. The main differences are the direction of TGI flow, either toward the carina (direct TGI) or toward the airway opening (reverse TGI). While controversy over direction and efficiency exists (5, 6, 17), reverse TGI delivery systems avoid potential mucosal damage caused by the jet stream and eliminate the increase in PEEP caused by the TGI flow with direct TGI. Our previous data with this TGI tube demonstrated CO2 elimination efficiency similar to direct TGI at both normal and inverse I:E ratios (6). The increase in PEEP with direct TGI can be large (> 5 cm H2O, depending on TGI flow) (6, 24) and requires monitoring and adjustment. However, monitoring is difficult because TGI flow must be interrupted during total PEEP measurement. Because we have never observed changes of more than 1 cm H2O at TGI flows up to 10 L/min with this prototype reverse TGI tube (reference 6 and current data), total PEEP monitoring and adjustment during reverse TGI with this tube may be unnecessary. Limitations

This study was performed in a sheep lung lavage model of ARDS, which may not reflect the same pathophysiology observed in human ARDS or in other ARDS models. Because our study focused on the change in Paw, our results are closely

VOL 160

1999

related to lung mechanics and the severity of the lung injury. We did not examine TGI flows higher than 10 L/min, as a result, the maximal efficiency of this approach cannot be assessed nor did we investigate variations in ventilator settings on the behavior of the release mechanism. In addition, the use of this TGI tube requires reintubation of critically ill patients, which may be problematic. In conclusion, continuous-flow TGI in combination with PCV1 allowed reductions in Paw and VT in healthy and lunginjured sheep while maintaining eucapnia. The configuration of the TGI delivery system maintained inspiratory and expiratory pressures constant because the exhalation valve was active during the inspiratory phase and the TGI flow was directed toward the airway opening. Acknowledgment : The authors thank Dräger Inc. for providing the Evita 4 ventilator and especially Dr. Weismann for developing the software, which allowed the total PEEP measurement during TGI.

References 1. Hickling, K. G., S. J. Henderson, and R. Jackson. 1990. Low mortality associated with low volume pressure limited ventilation with permissive hypercapnia in severe adult respiratory distress syndrome. Intensive Care Med. 16:372–377. 2. Amato, M. B., C. S. Barbas, D. M. Medeiros, R. B. Magaldi, G. P. Schettino, G. Lorenzi-Filho, R. A. Kairalla, D. Dieheinzelin, C. Munoz, R. Oliveria, T. Y. Takagoki, and C. R. R. Carvalho. 1998. Effect of a protective-ventilation strategy on mortality in the acute respiratory distress syndrome. N. Engl. J. Med. 338:347–354. 3. Slutsky, A. 1993. Mechanical ventilation: American College of Chest Physicians Consensus Conference. Chest 104:1833–1859. 4. Nahum, A., W. C. Burke, S. A. Ravenscraft, T. W. Marcy, A. B. Adams, P. S. Crooke, and J. J. Marini. 1992. Lung mechanics and gas exchange during pressure-control ventilation in dogs: augmentation of CO2 elimination by an intratracheal catheter. Am. Rev. Respir. Dis. 146:965–973. 5. Nahum, A., S. A. Ravenscraft, G. Nakos, W. Burke, A. Adams, T. W. Marcy, and J. J. Marini. 1992. Tracheal gas insufflation during pressurecontrol ventialtion: effect of catheter position, diameter, and flow rate. Am. Rev. Respir. Dis. 146:1411–1418. 6. Imanaka, H., M. Kirmse, H. Mang, D. Hess, and R. M. Kacmarek. 1999. Expiratory phase tracheal gas insufflation and pressure control ventilation in lung lavage sheep with permissive hypercapnia: effects of TGI flow direction and inspiratory time. Am. J. Respir. Crit. Care Med. 159:49–54. 7. Nahum, A., A. Chandra, J. Niknam, S. A. Ravenscraft, A. B. Adams, and J. J. Marini. 1995. Effect of tracheal gas insufflation on gas exchange in canine oleic acid–induced lung injury. Crit. Care Med. 23:348–356. 8. Nahum, A., S. A. Ravenscraft, G. Nakos, A. B. Adams, W. C. Burke, and J. J. Marini. 1993. Effect of catheter flow direction on CO 2 removal during tracheal gas insufflation in dogs. J. Appl. Physiol. 75:1238–1246. 9. Imanaka H., R. Kacmarek, R. Ritz, and D. Hess. 1996. Tracheal gas insufflation—pressure control versus volume control ventilation: a lung model study. Am. J. Respir. Crit. Care Med. 153:1019–1024. 10. Imanaka, H., R. M. Kacmarek, V. Riggi, R. Ritz, and D. Hess. 1998. Expiratory phase and volume-adjusted tracheal gas insufflation: a lung model study. Crit. Care Med. 26:939–946. 11. Nahum, A., R. S. Shapiro, S. A. Ravenscraft, A. B. Adams, and J. J. Marini. 1995. Efficacy of expiratory tracheal gas insufflation in a canine model of lung injury. Am. J. Respir. Crit. Care Med. 152:489–495. 12. Ravenscraft, S. A., W. C. Burke, A. Nahum, A. B. Adams, G. Nakos, T. W. Marcy, and J. J. Marini. 1993. Tracheal gas insufflation augments CO2 clearance during mechanical ventilation. Am. Rev. Respir. Dis. 148:345–351. 13. Burke, W. C., A. Nahum, S. A. Ravenscraft, G. Nakos, A. B. Adams, T. W. Marcy, and J. J. Marini. 1993. Modes of tracheal gas insufflation: comparison of continuous and phase-specific gas injection in normal dogs. Am. Rev. Respir. Dis. 148:562–568. 14. Lachmann, B., B. Robertson, and J. Vogel. 1980. In vivo lung lavage as an experimental model of the respiratory distress syndrome. Acta Anaesth. Scand. 24:231–236. 15. Nunn, J. F. 1993. Distribution of pulmonary ventilation and perfusion. In J. F. Nunn, editor. Applied Respiratory Physiology, 4th ed. Butterworths, Boston. 16. Feihl, F., and C. Perret. 1994. Permissive hypercapnia: how permissive

Kirmse, Fujino, Hromi, et al.: Pressure-release Tracheal Gas Insufflation should we be? Am. J. Respir. Crit. Care Med. 150:1722–1737. 17. Ravenscraft, S. A., R. S. Shapiro, A. Nahum, W. C. Burke, A. B. Adams, G. Nakos, and J. J. Marini. 1996. Tracheal gas insufflation: catheter effectiveness determined by expiratory flush volume. Am. J. Respir. Crit. Care Med. 153:1817–1824. 18. Sznajder, J. I., C. J. Becker, G. P. Crawford, and L. D. Wood. 1989. Combination of constant-flow and continuous positive-pressure ventilation in canine pulmonary edema. J. Appl. Physiol. 67:817–823. 19. Nakos, G., S. Zakinthinos, A. Kotanidou, H. Tsagaris, and C. Roussos. 1994. Tracheal gas insufflation reduces the tidal volume while PaCO2 is maintained constant. Intensive Care Med. 20:407–413. 20. Nahum, A., S. A. Ravenscraft, A. B. Adams, and J. J. Marini. 1995. Inspiratory tidal volume sparing effects of tracheal gas insufflation in dogs with oleic acid–induced lung injury. J. Crit. Care 10:115–121. 21. Gowski, D. T., E. Delgado, A. M. Miro, F. J. Tasota, L. A. Hoffman, and

1467 M. R. Pinsky. 1997. Tracheal gas insufflation during pressure control ventilation: effect of using a pressure relief valve. Crit. Care Med. 25: 145–152. 22. Okamoto, K., H. Kishi, H. Choi, and T. Sato. 1997. Combination of tracheal gas insufflation and airway pressure release ventilation. Chest 111:1366–1374. 23. Kolobow, T., T. Powers, S. Mandava, M. Aprigliano, A. Kawaguchi, K. Tsuno, and E. Mueller. 1994. Intratracheal pulmonary ventilation (ITPV): control of positive end-expiratory pressure at the level of the carina through the use of a novel ITPV catheter design. Anesth. Analg. 78:455–461. 24. Kalfon, P., G. S. Rao, L. Gallart, L. Puybasset, P. Coriat, and J. J. Rouby. 1997. Permissive hypercapnia with and without expiratory washout in patients with severe acute respiratory distress syndrome. Anesthesiology 87:6–17.