INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Despite improvements in outcome produced by antenatal corticosteroid therapy and exogenous surfactant therapy, the smallest newborns present a high rate of chronic lung disease. This poor outcome seems to be caused mainly by lung baro- or volutrauma (1), and pressure- and volume-reduction strategies for this patient population are now being actively studied. To reduce tidal volume (VT), high-frequency oscillatory ventilation (HFOV) has been proposed for many years (4), but its superiority over conventional ventilation remains uncertain, even 10 yr after the first study by the HiFi group (7). Reduction of VT during conventional ventilation, also referred to as permissive hypercapnia, has been mainly described in adult patients (10), and little documentation can be found to confirm its benefit in premature newborns, in whom hypercapnia and acidosis may be of greater concern than in adults (16). Continuous tracheal gas insufflation (CTGI) is an interesting option, since it can significantly decrease ventilation pressures (17). CTGI is primarily based on the notion that instrumental dead space becomes a major determinant of minute ventilation (E) when very small tidal volumes are used. It has been previously shown that CTGI could be specifically adapted for premature newborns, and was safe and feasible for long-term use (22, 23). However, its efficiency in providing pressure reduction throughout the period of ventilatory support (i.e., until weaning), has not yet been demonstrated. In addition, the effect of CTGI on oxygenation required study in a controlled fashion because reduction in mean airway pressure may result in progressive alveolar derecruitment (13). To address these two questions and to examine the influence of CTGI on outcome, we designed a prospective, randomized study comparing pressure-controlled conventional ventilation with and without CTGI.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The study was conducted between January 1997 and February 1999 (with a temporary interruption from mid-April to mid-June 1997 because of closure of the neonatal intensive care unit). The study protocol was approved by the ethics commitee of the Créteil-Hôpital Henri Mondor area. After informed consent was obtained from both parents, newborns of gestational age, < 30 wk, treated with natural surfactant (Curosurf 150 mg/kg; Chiesi, Parma, Italy) and ventilated for hyaline membrane disease, were randomized, in the 6 h after birth, to receive CTGI associated with conventional pressure-controlled ventilation or pressure-controlled ventilation alone (controls). Patients with ischemic encephalopathy, refractory hypotension, lethal malformations, or congenital malformations with pulmonary consequences were excluded from the study. Mechanical ventilation was provided with a pressure-limited ventilator (Babylog 8000; Drager, Lubeck, Germany) and was flow-triggered for intermittent synchronization. The initial ventilatory settings were standarized for all infants, and included a positive end-expiratory pressure (PEEP) set at 4 cm H₂O, a respiratory rate (RR) of 60 cycles/min (cpm), an inspiratory time of 0.35 s, and continuous flow at 10 L/min. The fraction of inspired oxygen (FI_O2) was adjusted to keep the oxygen saturation (Sa_O2) in the range of 90 to 95%, and the peak inspiratory pressure (Ppi) was adjusted to keep the transcutaneous Pa_CO2 (tcPa_CO2) in the range of 40 to 46 mm Hg. All of the newborns enrolled in the study were initially intubated with a 2.7-mm I.D. multichannel endotracheal tube (ETT) specially adapted to deliver gas at the tracheal end of the tube (No. 6501.25; Vygon, Ecouen, France).

CTGI was provided as described in previous studies (22, 23), by an ancillary flow injected through six capillary tubes within the wall of the ETT and running parallel to its lumen, a single connection being used for the input of oxygenated, heated, and humidified gas (Figure 1). A 0.5 L/min flow rate was used and was provided by a silent pump. CTGI circuit pressure and tracheal pressure were continuously monitored, and a control system, connected to an audible alarm, was included to automatically disconnect the pump if the CTGI circuit pressure or the tracheal pressure increased by 10%. The efficacy of the alarm and disconnection system had been checked in a bench test. We found that the delay before the flow was stopped was always less than 500 ms. The only modification induced by a 0.5 L/min CTGI flow in the respiratory system is a continuous increase in distal tracheal pressure of 0.8 cm H₂O, which cannot be detected at the site of proximal airway pressure measurement (23).

View larger version (43K): [in this window] [in a new window]   Figure 1.   Schema of the continuous tracheal gas insufflation (CTGI) device: a pump located in the incubator draws a flow of 0.5 L/min from the inspiratory circuit downstream to the heater-humidifier and feeds the gases into the CTGI circuit. The CTGI pump is pressure limited by a monitor with a cutoff system. As shown in the endotracheal tube (ETT) cross-section, six capillaries (black) are used for CTGI and two others (gray) are used for tracheal pressure monitoring (A) and exogenous surfactant treatment (B).

Randomization and Treatment Group

Randomization was achieved through the use of sealed envelopes. The ventilatory treatment was continued if the oxygenation index (OI = mean Paw × FI_O2 × 100/Pa_O2) remained within the range of 2 to 12. An OI above 12 was used to define severe hypoxemia and to switch to HFOV (OHF1; Dufour, Villeneuve d'Ascq, France). The infants were weaned from ventilatory support when the ventilation rate was below 30 cpm and the OI was below 2 mbar/mm Hg. For the CTGI group, CTGI was removed 30 min before assessment of weaning criteria. Because of the 0.8 cm H₂O increment in lung pressure produced by the 0.5 L/min CTGI flow, external PEEP was decreased by this amount in the CTGI group to ensure the same PEEP level in the trachea and the lungs in both groups. Corticosteroid treatment was initiated in the following circumstances: during the first week if the OI rose above 20 mbar/mm Hg; during the second week if the OI exceeded 8 mbar/mm Hg; and after 2 wk when extubation remained impossible.

Standard Monitoring

All patients were monitored with a multiparameter monitoring system providing an electrocardiogram, respirogram, tcPa_CO2 and tcPa_O2, pulse oximetry (Sa_O2), blood pressure by oscillometry, and ventilatory parameters transferred directly from the ventilator to a monitor (Viridia CMS; Hewlett-Packard, Boeblingen, Germany). Because of large breath-to-breath fluctations resulting from CTGI superimposed on spontaneous breathing activity, integrated flow data, VT, and E could not always be very accurately recorded. Mean VT and E values are therefore given only for the time of randomization.

Data Collection

During ventilatory support, we compared the driving pressure (P = (Ppi  PEEP), the respiratory index (tcPa_O2/FI_O2), tcPa_O2, tcPa_CO2, Sa_O2, heart rate and spontaneous breathing rate, ventilatory settings (inspiratory time [TI], ventilation rate, continuous flow rate, mean airway pressure), VT, and E. Data were collected before and at the time of randomization, at 2, 4, 6, 8, 14, and 24 h, and then daily until Day 28. The two main objectives were to compare the pressure course in both study groups and to follow the oxygenation status with use of CTGI. In addition, the first date of extubation (OI  2 mbar/mm Hg and ventilation rate < 30 cpm), and the final date of successful extubation, defined as an extubation lasting more than 7 d without failure, were recorded, as were the total duration of mechanical ventilation and the patient's respiratory and general status at 28 d of life and at 36 and 42 wk gestational age. Prolonged need for oxygen usually defines chronic lung disease. The need for oxygen was therefore assessed in both groups at these same stages.

Statistics

Comparisons of the CTGI and control groups were performed with the Mann-Whitney U test for continuous data, and with chi-square tests for categorical data or Fisher exact test when necessary.

For the first extubation time, a Kaplan-Meier curve was constructed and the two groups were compared with the log-rank test. Duration of endotracheal ventilation before the definitive extubation time was analyzed with the Mann-Whitney U test.

Differences were considered significant for values of p < 0.05. For analysis of repeated measures, such as P, Bonferroni's correction was applied to modify the p-value threshold for significance. Statistical analyses were done with Statview software version F4.5 for Macintosh computers (Abacus Concepts, Berkeley, CA).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Among 60 preterm neonates presenting with hyaline membrane disease, 41 consecutive newborns requiring mechanical ventilation and surfactant therapy and meeting all inclusion criteria were randomized in the study. Seven newborns were excluded after randomization. In three of these seven cases, CTGI could not be used because of technical problems (two because of leaks in the CTGI circuit and insignificant CTGI flow and one because of a wrong CTGI circuit setup). For a pair of twins, the parents withdrew their consent after randomization despite a prior protocol agreement. In one case, randomization was started without parental agreement, and the protocol was not performed. The last exclusion involved a newborn with maternofetal infection who required HFOV at the time of randomization, when he developed a problem that met the study exclusion criteria (OI > 12 mbar/mm Hg and hemodynamic collapse).

Analysis was thus done on data from the 34 newborns who did undergo the study treatments: 15 in the CTGI group and 19 in the control group, with a mean gestational age of 27.8 ± 1.3 wk (mean ± SD) and a mean weight of 965 ± 202 g. Randomization was done at 226 ± 81 min of life, at 50 ± 42 min after the first dose of surfactant. The two groups were not different for gestational age, weight, age at randomization, ventilatory settings at the study inclusion time, hospital-born-to-outborn ratio, or antenal steroid therapy (Table 1).

In the CTGI group, CTGI was applied throughout the intubation period, from the time of randomization. The longest duration of CTGI administration in the study was sixteen consecutive days.

Respiratory parameters were analyzed for the first 4 d. After this period, the number of ventilated patients became too small to make comparisons meaningful. As shown in Figure 2, P (Ppi  PEEP) was lower in the CTGI group at each time point, and the difference was significant at p < 0.005 for Hours 4, 14, 24, 48, and 96. On average throughout the first 4 d, the mean P value was 10.8 ± 2.3 cm H₂O in the CTGI group and 14.5 ± 2.2 cm H₂O in the control group. The mean reduction in P with CTGI was 29%, from 18% at Hour 8 to 35% at Day 4. There was no difference between the two study groups in TI, ventilator frequency, or PEEP at any time,

View larger version (14K): [in this window] [in a new window]   Figure 2.   Effect of CTGI on respiratory status over the first 4 d after randomization. Comparison of the CTGI group (closed circles) and control group (open circles) for the difference between Ppi and PEEP (P). (A); tcPaCO2 (B); and tcPaO2/FIO2 (C ) CTGI allowed a significant reduction of P without alteration of oxygenation status or tcPaCO2. (D) Number of extubated newborns (lower panel ) and those treated with high-frequency oscillatory ventilation (HFOV) (upper panel ) at each time point. In the CTGI group, extubation occurred sooner, and there was more switching to HFOV in the control group during this time.

There was no statistical difference between the two study groups in tcPa_CO2 at any of the study time points. During the first 4 d, mean tcPa_CO2 was 41.8 ± 3.0 mm Hg for the CTGI group and 42.2 ± 3.8 mm Hg for the control group.

There was also no statistical difference in the tcPa_O2-to-FI_O2 ratio for the two groups at any study time point. During the first 4 d, the mean tcPa_O2/FI_O2 value was 245 ± 29 in the CTGI group and 261 ± 46 mm Hg in the control group. The same results were found for tcPa_O2 and FI_O2 evaluated separately.

The Kaplan-Meier curve applied to extubation time is shown in Figure 3, in which it is seen that the two groups differed in this measure (p < 0.05 with the log-rank test). The first extubation occurred significantly sooner in the CTGI group, at a median of 2.1 d (25th to 75th quartiles: 0.9 to 10.5 d) for the CTGI group versus a median of 8.4 d (25th to 75th quartiles: 2.6 to 20.7 d) for the control group (Figure 3). With regard to the successful extubation time (no need for reintubation for the next 7 d), the median was 5.2 d (25th to 75th quartiles: 1.5 to 13.3 d) in the CTGI group versus 16.9 d (25th to 75th quartiles: 8.3 to 22.4 d) in the control group. As a result, the total duration of endotracheal ventilation for survivors was shorter in the CTGI group, at 3.6 (1.5 to 12.0 d) (mean and 25th and 75th quartiles) versus 15.6 (7.9 to 22.2) d (p = 0.012), as shown in Figure 4.

View larger version (16K): [in this window] [in a new window]   Figure 3.   Kaplan-Meier curve applied to the first extubation time for the CTGI group (closed circles) and for the control group (open circles). The log-rank test indicated a significant difference between the two groups (p < 0.05).

View larger version (26K): [in this window] [in a new window]   Figure 4.   Comparison of the duration of mechanical ventilation for the CTGI group (white box) and the control group (dotted box), with the 10th, 25th, 50th, 75th and 90th percentiles indicated for each group. The total number of days on ventilation (median and 25th to 75th quartiles) was 3.6 (1.5-12.0) d for the CTGI group versus 15.6, (7.9- 22.2) d for the control group (p = 0.012, Mann-Whitney U test).

There was no difference between the two groups in outcome or complications (Table 2). Five deaths occurred, two at Days 5 and 7 in the control group, due to hemodynamic and respiratory failure, and three after 28 d, including two deaths from cystic periventricular leukomalacia in the control group and one death from postnatal infection with respiratory failure in the CTGI group.

Indices of chronic lung disease did not differ in the two groups. Few newborns needed O₂ at 28 d in either group (Table 2). Eight of the 34 newborns (24%) who completed the study protocol reached the criteria for HFOV.

One newborn in the CTGI group presented an ETT obstruction requiring two ETT replacements, the first during CTGI ventilation and the second during the course of HFOV. We could not identify any failure in the humification sytem, and the CTGI security device was efficient.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The present study was the first long-term, controlled randomized study to evaluate the efficacy of CTGI in human patients. It confirmed that CTGI associated with conventional ventilation is efficient for decreasing the peak inflation pressure by almost one third over a prolonged period of ventilation. Applied to premature newborns, this technique induced a significant reduction in the duration of ventilatory assistance. Respiratory and general management were designed to be entirely identical in both of the study groups, allowing firm conclusions about the effects of CTGI. Moreover, in several patients, Ppi could be decreased dramatically, although the software used in the ventilator did not allow a Ppi below 10 cm H₂O.

The efficacy of lung-protective strategies is debated, and different adult studies have reached contradictory conclusions about it (15, 24, 25). In addition, the safest level of Pa_CO2 in neonates is a matter of debate. In contrast to permissive hypercapnia, our approach is not based on a deliberate reduction of VT (16) but is based on analysis of the cause of a high dead space volume VD-to-VT ratio due to inappropriately high instrumental dead space (26). In this study, the target for Pa_CO2 was the upper limit of normocapnia (40 to 46 mm Hg). If permissive hypercapnia became the rule for neonates, CTGI could be used with even greater efficacy because reducing the VT increases the efficacy of dead space washout (23). Before CTGI, the approach used in the neonatal intensive care unit was to cut the ETT, which, however, only marginally resolved the dead space problem, whereas it made ventilation of the infant less secure and comfortable. The use of CTGI entirely solves the instrumental dead space problem, and allows use of a lower but still efficient VT. Several animal studies have shown that tracheal gas insufflation (TGI) decreases dead space, VT, and E (21, 27), and may decrease the work of breathing (18, 30). Most of these studies used a catheter for providing TGI, and volume-controlled ventilation was the primary mode of ventilation. In neonates, pressure-limited ventilation is the primary mode of ventilation. It is noteworthy that a CTGI at 0.5 L/min used with neonatal continuous-flow ventilators does not modify minute E or VT (data not shown), in contrast to data reported with volume-cycled but also pressure-limited ventilators for adults (31). In addition, a specific multichannel ETT (32, 33) was used in our study because the small size of the ETT lumen in neonates forbids the insertion of an additional catheter. Another approach, based on aspiration of expired CO₂, has recently been proposed (34). For the same reasons of tube size, the aspiration channel, to be efficient, would be too large in neonates. For neonatal application, Kolobow and coworkers developed the technique of intratracheal pulmonary ventilation, which permits a full bypass of instrumental dead space through use of a catheter introduced into the ETT lumen (35). There are very few human studies in the literature of TGI (17, 21, 30, 36), and no randomized or long-term trials. With regard to preterm infants, we had previously shown that CTGI allowed a 30% reduction in P during a 30-min test, and was suitable for long-term use (22). In a second study, we had shown that the efficacy of CTGI was correlated with Pa_CO2 values before CTGI, and with the instrumental VD-to-VT ratio (23).

By decreasing VT as well as mean airway pressure, CTGI might be expected to cause alveolar instability and poor alveolar recruitment, as suggested by others (4, 13, 37). Conversely, a high VD-to-VT ratio during conventional ventilation implies lung overinflation. In this context, we found that CTGI had a precisely corrective effect and restored an appropriate VT to maintain a normal Pa_CO2. Whatever the hypothesis, it was important to check the effect of CTGI on oxygenation status. In this study, the ratio of tcPa_O2 to FI_O2 remained identical in both study groups.

By increasing expiratory flow, CTGI could be suspected to amplify expiratory resistance. Because the mean expiratory flow in neonates is low, and an additional CTGI flow of 0.5 L/ min is very low, we did not observe a substantial impact of CTGI in this flow range on expiratory resistance. The shape of the respiratory flow curve was not modified during CTGI (data not shown), suggesting that CTGI does not modify either resistance or compliance of the respiratory system. Moreover, since CTGI allows a reduction in P, expiratory resistance may be reduced.

Both groups of newborns in the study were weaned from ventilation according to the same criteria. In particular, for calculation of the OI, CTGI was interrupted for 30 min before recording of the necessary values and performance of the calculations. It could be argued that our criteria for weaning of infants from ventilation were rather difficult to meet. Very few criteria for extubation in premature neonates, however, have been published in the literature (38). The protocol used for weaning was similar to our usual practice. The thresholds used may have influenced the results, in view of the long duration of ventilation in the control group, but conversely could explain the relatively low rate of reintubation in both groups (15%). Even in this case, however, the difference in P between the two groups tended to increase with time for the infants who remained intubated, which may suggest the hypothesis of a lung-protective effect of CTGI in the early hours after birth. Introduction of CTGI immediately after birth would therefore be interesting to test.

The sample size of the study was not large enough to detect any impact of CTGI on long-term outcome. It is interesting to note, however, the reduced use of inotropic drugs in the CTGI group. The reduced need for hemodynamic support could be expected on the basis of the decrease in intrathoracic pressure offered by the low-pressure ventilation strategy entailed in CTGI. The reduced need for ventilatory support could suggest a protective effect of CTGI. However, a larger study, available only through a multicenter randomized trial, is required to confirm this hypothesis.

One of our patients plugged his ETT with mucus during the CTGI procedure. It is difficult to implicate CTGI in this complication because the same patient again plugged his ETT during HFOV, and in our previous studies this problem never occurred. Nevertheless, this event reinforces the importance of an adequate humidification of CTGI gases, associated with well-designed security features. In this latter regard, the security system used in our study was efficient and prevented hyperinflation of the lungs.

CTGI seems to be helpful in preterm neonates with hyaline membrane disease. To be used routinely, it requires a specific monitoring and safety system to avoid potential side effects. Previous bench tests indicated that the optimal CTGI flow with the present system was 0.5 L/min in neonates (23). At this flow rate, there was no modification of humidity or temperature of the gases participating in alveolar ventilation, and E remained identical in the two study groups (23). Additionally, CTGI may slightly modify flow monitoring and respirator trigger efficiency (39), but these limitations could be easily corrected through software adaptations. The availability of such a system would permit manufacturers to integrate this option into their currently licensed ventilators.