INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The respiratory role of the larynx is of major importance for successful transition from fluid lung to aerated lung. During fetal life, control of glottic aperture regulates liquid movements between trachea and amniotic cavity and consequently is a major component of lung growth regulation (1). At birth, after a first inspiration, the glottis of the newborn is actively closed during the first forceful expirations. This mechanism allows establishment of functional residual capacity (FRC) (2). Thereafter, during the first hours of life, active glottic expiratory closure exerts an expiratory airflow braking which maintains an optimal FRC and probably contributes to the overall process of lymphatic drainage of lung fluid (3). Clinical importance of expiratory airflow braking is obvious in preterm infants in whom bypassing the larynx by endotracheal intubation precipitates blood gas deterioration (4). Moreover, from the onset of the first breath at birth, active laryngeal opening must be perfectly coordinated with inspiratory efforts. This is accomplished through glottic dilator muscle inspiratory contraction, thereby decreasing upper airway resistance and inspiratory work. Hence, precise control of glottic aperture within each phase of respiratory cycle attests to an essential role to the larynx in the first hours and days of life.

Respiratory rhythm instability, manifested by recurrent apneas, both isolated and during periodic breathing, is characteristic of early life in mammals. While fetuses in utero commonly close their glottis throughout 15- to 20-min apneas (5), it is still unknown whether the glottis is open or closed during apneas in the newborn. The fact that investigators endoscopically observed glottic closure throughout spontaneous central apneas in preterm infants (6) prompted us to examine the relationship between glottic control and central apnea. In a series of experiments, we recently showed that continuous thyroarytenoid (TA) muscle (a glottic constrictor) electromyographic (EMG) activity was consistently present throughout artificially induced central apneas in nonsedated term lambs in the same manner as the fetus (7, 8). Physiological relevance of these unique results, unrelated to active glottic closure triggered by stimulation of laryngeal receptors (i.e., the laryngeal chemoreflex), remained unknown, however.

Using repeated polysomnographic recordings, the present experiments were aimed at studying whether active glottic closure is present during spontaneous apneas in nonsedated preterm lambs with intact upper airways. Presence and patterns of active glottic constriction, including relation with age and state of consciousness, could hence be analyzed during ventilatory and apneic periods.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Fourteen preterm lambs (postconceptional age 131.4 ± 1.1 d, range 129 to 132 d) were involved in this study (normal term 147 d). Their mean weight was 2.7 ± 0.5 kg (range 2.1 to 3.5 kg). The protocol of the study was approved by the ethics committee of the Université de Sherbrooke for animal care and experimentation.

Preterm Lamb Model

Antenatal lung maturation. Pregnancies were dated accurately by means of single mating after induction of estrus with vaginal sponges containing 60 mg of medroxyprogesterone acetate. Fetal ultrasound examination of ewes 50 d after mating was performed to verify pregnancy. To prevent respiratory distress syndrome in preterm lambs at birth, fetal lung maturation was accelerated by administering one dose of betamethasone and several doses of thyrotropin-releasing hormone (TRH; Relefact; Hoechst-Marion-Roussel, Kansas City, MO) to the ewe as follows (9): at 48 h before delivery, 0.5 mg/kg of betamethasone was given by intramuscular injection and a first dose of 400 µg of TRH was given by slow intravenous infusion (2 min). Three additional doses of 400 µg of TRH were subsequently infused at 36, 24, and 12 h before delivery.

Preterm delivery. Twelve hours before delivery, a prostaglandin gel (PGE₂) was intravaginally placed to accelerate uterine maturation. On the day of delivery, a continuous infusion of oxytocin was given intravenously to the ewe to activate preterm labor. The infusion was started at a rate of 2 mUI/mn and subsequently increased every 20 min up to a maximal rate of 40 mUI/mn. Labor evolution was monitored by means of regular physical examination. Finally each lamb was born vaginally without help except in case of adverse presentation.

Postdelivery care. Immediately after delivery, each lamb was dried with warmed towels and weighed. Following clinical examination, rectal temperature (Mon-a-Therm; Mallinckrodt, Pointe Claire, PQ, Canada) and transcutaneous oxygen saturation (Nonin 8500; Nonin Medical Inc., Plymouth, MN) were regularly verified. The newborn lamb was left with its mother in normal conditions. In case of hypothermia (< 38.5° C) and/or hypoglycemia (< 2.3 mmol/L, BM test; Boehringer Ingelheim Ltd, Burlington, ON, Canada) because of insufficient spontaneous feedings, the lamb was warmed (at 39° C) in an incubator and bottle-fed with mother's milk. In case of respiratory distress syndrome, standard care in use in our institution for human newborns, including endotracheal ventilatory support (Babylog-1; Dräger, Lübeck, Germany) and exogenous surfactant replacement (Survanta; Abbott Laboratories Ltd, Montreal, Canada), was implemented.

Surgical Preparation

Surgery was performed 1 to 3 d after birth under general anesthesia (Fluothan 2 to 3% + N₂O 30%). Atropine sulfate (150 µg/kg subcutaneous) was systematically given preoperatively with ketamine 10 mg/ kg (Ketaset; Ayerst, Montreal, Canada) and midazolam 100 µg/kg (Versed; Hoffman-La Roche, Mississauga, ON, Canada). The same dosage of atropine sulfate was repeated once during the procedure.

Bipolar enameled chrome wire electrodes (0.1 mm diameter, Chromel; GTSM, Castelnaudary, France) were inserted into the TA (a glottic constrictor), inferior pharyngeal constrictor (IPC), and diaphragm muscles for EMG activity recording, as previously reported (8, 10). Briefly, an EMG electrode was sewn under direct vision into both TA muscles through a small window made on each side of the thyroid cartilage. Then, the larynx was cautiously turned to visualize the right IPC and an electrode was inserted perpendicularly to the muscular fibers near the posterior margin of thyroid cartilage. For the diaphragm, two electrodes were sewn in the costal portion of the right hemidiaphragm after a small thoracotomy. Correct positioning of the electrodes in the muscles was always verified at autopsy. Three platinum needle electrodes (F-E₂; Grass Instrument Company, Quincy, MA) were inserted under the periosteum in the frontal and parietal bones. Two needles were used for electroencephalogram (EEG) recording, the remaining needle acting as a ground. Two other platinum needle electrodes were placed into the outer and upper part, and the inner and lower part of the right eye socket for electrooculogram (EOG) recording. Nasal airflow was recorded from a thermocouple wire (iron/constantan, type J; Omega Engineering, Stamford, CO) glued to the side of one nostril (Rapid Adhesive X 60; Hottingen Badwin Messtechnik, Darmstadt, Germany). Finally, two cup-electrodes (Neotrode; ConMed Andover Medical, Haverhill, MA) were placed subcutaneously in the anterior chest wall for electrocardiogram (ECG) recording. The leads of each electrode were subcutaneously tunneled to exit on the back of the lambs.

Postoperative care included observation in incubator and continuous monitoring of rectal temperature and Sa_O2. In case of postoperative apnea, caffeine and doxapram were used when stimulations were not sufficient. After complete recovery, lambs were housed with their mother in our animal quarters until recordings and received ampicillin and gentamicin daily for 10 d.

Measurement Apparatus

In order to obtain data from prolonged recordings (with periods of wakefulness and sleep) under more natural conditions, we recently designed a radiotelemetry device which enables us to study free moving lambs. The transmitter (3.5 × 2.5 × 0.8 cm) used in the present study was made of eight differential channels designed for EMG (four channels), nasal flow, ECG, EOG, and EEG, each channel having its own differential preamplifier and high-pass filter. The eight signals were transmitted on the same radio frequency using temporal multiplexing. The resulting signal, pulse amplitude modulation/time division multiplex (PAM/TDM), was demultiplexed by the receiver to rebuild the eight original signals. For more flexibility, each channel possessed its own variable gain amplifier and polarity selector at the receiver level. Finally, the eight signals were sent from the receiver to the acquisition system. The EMG signals were amplified and bandpass filtered 30 to 1,000 Hz (P511 AC preamplifier and 7 DADC drive amplifier; Grass). The raw EMG signals were then fed to a moving time averager (Department of Electronics, Faculty of Medicine, Université de Sherbrooke) with a time constant of 100 ms. Raw and integrated EMG, EOG, EEG, ECG, and nasal flow signals were recorded on a 10-channel polygraph (Model 7D; Grass). During the study, the transmitter was connected to the leads of each electrode on the back of the lamb and the receiver was placed in a different room for data acquisition.

Design of the Study

Each lamb was studied 3 times a week from birth to at least normal term. The first recording was made 48 h after surgery, with duration of recordings between 3 to 8 h. The nonsedated lamb had only the transmitter attached to its back, a rectal temperature sensor, and a pulse oximetry sensor on one shaved ear. The recording area was a large neonatal plexiglass incubator with servo-controlled temperature and continuous airflow (10 L/min) to avoid rebreathing. Lambs were continuously monitored using a video camera placed in the recording room. Data acquisition was performed in an adjacent room, where an observer noted all events occurring during the recordings.

Data Analysis

Variables: definitions and measurements. All signals were carefully observed and analyzed throughout the recordings according to state of consciousness of the lamb. All results of studied variables were given for all lambs as a whole. Variables recorded for each lamb are described below.

State of consciousness. Usual electrophysiological and behavioral criteria were used to define state of consciousness. We identified: (1) wakefulness from the presence of high-frequency, low-voltage EEG pattern with open eyes and occasional body movements; (2) non- rapid eye movement (nREM) sleep from slow-frequency, high-voltage EEG pattern with closed eyes, absence of rapid eye movements (REM), and body movements; and (3) REM sleep from the presence of high-frequency, low-voltage EEG pattern, with bursts of REM and body twitches. Arousal from nREM sleep was characterized by sudden disappearance of high-amplitude waves in the EEG. Arousal from REM sleep was recognized by direct observation of the lamb and disappearance of intense EOG activity.

Apneas and periodic breathing. Apneas were defined as absence of nasal airflow for at least 3 s and were classified as central, mixed, or obstructive from diaphragm EMG (respectively absent or present during part or throughout apnea). Periodic breathing (PB) was defined as alternating series of contiguous breaths and apneas ( 3 s) or hypopneas. The apnea index (the number of apneas per hour of sleep in each state) and duration of each apnea, as well as the number of oscillations and total duration of each PB episode were calculated. Heart rate and transcutaneous Sa_O2 during apnea and PB were analyzed. Oxygen desaturation was defined as a 10% or greater fall in Sa_O2 or a fall under 90%. Bradycardia was defined as a 30% or greater fall in heart rate. We also noted whether apneas were preceded by an event such as a swallowing, cough, or sigh or if they occurred suddenly.

TA and IPC EMG. TA muscle and IPC EMG were analyzed throughout the recordings. Apart from apneas or PB episodes, duration of ventilation with or without expiratory TA and IPC EMG was calculated. During apneas (either isolated or within PB) we noted if TA or IPC EMG were present, and whether they were continuous throughout the apneas. Our study design, however, did not enable us to ascertain whether the glottis was completely or only partially closed when TA EMG was present and airflow nil. Consequently, the term "glottic closure" was used in its broader sense, without inferring if glottic closure was complete or partial. However, the nasal airflow signal was highly amplified to allow the use of the cardiac artefact as an indicator of upper airway patency during apneas (11).

Research queries. The first outcome in our study was the presence of TA EMG and IPC EMG during apneas, being isolated or in PB. Secondary outcomes in this novel animal model included characteristics of apneas observed (index, duration) as well as their consequences on Sa_O2 and heart rate, and TA EMG and IPC EMG during the ventilatory periods. Two potential predictors of all variables to be controlled were the state of consciousness of the lamb as well as its maturation (age). Specific questions were divided into: (1) TA EMG and IPC EMG activities during ventilatory periods, including the description of TA and IPC EMG within breathing cycles, the effects of states of consciousness and maturation with respect to the percentage of breaths with TA EMG as well as with IPC EMG, the relation between presence of IPC EMG and presence of TA EMG; (2) characteristics of apneas and PB, including the effects of state of consciousness and maturation on apnea index and apnea duration, and on the percentage of time spent in PB; (3) TA EMG and IPC EMG during apneas and periodic breathing, including the effects of states of consciousness and maturation on the number of apneas with continuous TA EMG as well with IPC EMG; and (4) O₂ desaturation and bradycardia events, including the effects of states of consciousness and maturation as well as apnea duration and TA EMG pattern (continuous or not) on the presence of O₂ desaturation and bradycardia events.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Fourteen preterm lambs were involved in the study. Eight survived without postdelivery intensive care and underwent the surgical procedure. Age at surgery was 46 ± 29 h (range 20 to 96 h) after birth. Numerous apneas were observed in the hours following surgery in all 8 preterm lambs and rendered continuous observation by a neonatologist necessary during 12 to 24 h postoperatively. These apneas were not recorded and consequently were not taken into account in our analysis. Early postoperative apneas were frequently prolonged and accompanied by desaturation and bradycardia; they required stimulations and treatment by caffeine (eight lambs) and intravenous infusion of doxapram (seven lambs). Therapeutic interventions were systematically limited to the first 24 h after surgery. One of the eight instrumented lambs was found dead 48 h postoperatively, before any recording had been performed, presumably from prolonged apnea (no significant findings at autopsy).

States of Consciousness

Forty-two recordings were performed in the seven surviving lambs, with a mean number of 6 ± 2.1 recordings per lamb. Age at first recording was 4 ± 1.5 d and at last recording 17 ± 7.5 d (postconceptional age was respectively 135 ± 0.7 d and 149 ± 7.9 d). Total duration of recordings for each state of consciousness is reported in Table 1. Mean durations per recording of REM sleep, nREM sleep, and wakefulness were respectively 42 ± 23, 72 ± 30, and 67 ± 48 min. Mean ratio of sleep duration versus wakefulness duration was 2.65 ± 2.24 and did not decrease during the study period (p = 0.92). Mean ratio of REM sleep duration versus total sleep duration (REM + nREM sleep) was 0.36 ± 0.12; this ratio significantly decreased during the first 3 w after birth (p = 0.004).

Thyroarytenoid and IPC EMG During Ventilatory Periods

When present during steady ventilation, TA EMG was noted to be in phase with onset of expiration and state-dependent. No TA EMG was noted during inspiration in any of the lambs. Generally, TA expiratory EMG began after transition between different states of consciousness and gradually decreased toward disappearance afterwards, but could reappear after swallowing, cough, or head movements. Furthermore, TA expiratory EMG rapidly disappeared when respiratory rate increased. During wakefulness, TA phasic expiratory EMG was present during 51 ± 26% of breaths, typically in the minutes following transition from nREM sleep, when lambs were still drowsy (Figure 1). TA muscle phasic expiratory EMG disappeared upon complete awakening. During nREM sleep, TA expiratory EMG was present during 62 ± 30% of breaths and of regular amplitude. The percentage of breaths with TA expiratory EMG was significantly lower during REM sleep (10 ± 18%) as compared with both nREM sleep (62 ± 30%) and wakefulness (51 ± 26%) (p < 0.0001). Furthermore, TA expiratory EMG during REM sleep was quite irregular (both in amplitude and occurrence) compared with nREM sleep. Finally, there was no significant relationship between presence of TA expiratory EMG and age within the study period, regardless of the state of consciousness (p = 0.78).

View larger version (38K): [in this window] [in a new window]   Figure 1.   TA and IPC EMG activity during nREM sleep in a nonsedated preterm lamb (postconceptional age = 140 d). From top to bottom: Thyroarytenoid: raw and moving time-averaged TA muscle electrical activity; IPC: raw and moving time-averaged electrical activity of the IPC muscle; Respitrace: respiratory volume measured by respiratory inductive plethysmography (sum signal); Thermistor: nasal airflow. Note TA expiratory EMG and the characteristic pattern of expiratory IPC EMG, i.e., with an initial and a final peak.

When present, the shape of the integrated IPC phasic expiratory EMG had the previously reported distinctive pattern (Figure 1) composed of an initial and a final peak (10). While the presence of IPC phasic expiratory EMG was significantly related to presence of TA EMG, irrespective of the state of consciousness (p < 0.0001), IPC expiratory EMG was less often observed than TA. Percentage of breaths with expiratory IPC EMG was significantly decreased during REM sleep (13 ± 22%) as compared with nREM sleep (51 ± 33%, p < 0.0001) and wakefulness (38 ± 28%, p = 0.001). No significant relationship between presence of IPC EMG and age could be found within the study period, regardless of the state of consciousness (p = 0.54).

Characteristics of Apneas

Central apneas. Most (96.7%) of the 2,088 apneas recorded were central. Characteristics of apneas for each state of consciousness are reported in Table 1. The central apnea index was significantly higher in nREM sleep (23.5 ± 47.2 h¹) compared with both REM sleep (10 ± 13.5 h¹, p = 0.005) and wakefulness (11.2 ± 31.9 h¹, p = 0.001). However, regardless of the state of consciousness, there was no significant relationship between age and apnea index (p = 0.38). Apneas shorter than 6 s were significantly more frequent than apneas longer than 6 s in wakefulness (n = 339 and 210 respectively, p = 0.007), but not in nREM (n = 546 and 662, p = 0.77) nor in REM sleep (n = 179 and 152, p = 0.33).

A prolonged, life-threatening apnea was recorded in one preterm lamb at a postconceptional age of 136 d during nREM sleep (Figure 2) and required vigorous stimulations and bagging with pure O₂ for a few seconds.

View larger version (40K): [in this window] [in a new window]   Figure 2.   Prolonged, life-threatening central apnea occurring spontaneously in a nonsedated preterm lamb (postconceptional age = 136 d) during nREM sleep. See Figure 1 for abbreviations. Diaphragm: raw and moving time-averaged diaphragm; EOG: eye movements. Spontaneous prolonged central apnea (total duration = 55 s) is accompanied by continuous TA EMG and extreme desaturation (to "zero" reading). Vigorous stimulation of the lamb and O2 breathing (arrow) lead to breathing resumption.

Mixed apneas. Only 68 mixed apneas and no obstructive apneas were recorded during the entire study. Ten mixed apneas were observed during wakefulness, 16 during nREM sleep and 42 during REM sleep. Mixed apnea index was significantly higher in REM sleep (1.3 ± 3.1 h¹) compared with wakefulness (0.3 ± 0.8 h¹, p = 0.02), but not when compared with nREM sleep (0.4 ± 0.9 h¹, p = 0.57). However, irrespective of the state of consciousness, there was no significant relationship between age and mixed apnea index (p = 0.54).

Mean duration of mixed apnea was 6.7 ± 2.7 s with only one inspiratory effort per apnea. Mixed apnea duration was significantly lower in wakefulness (5.4 ± 1.5 s) compared with both nREM (6.8 ± 2.4 s, p = 0.001) and REM (6.9 ± 3 s, p = 0.04) sleep. However, mixed apnea duration was positively related to age within the study period (p < 0.0001), even when controlling for state of consciousness (p = 0.0001). Finally, when controlling for age, only the difference between wakefulness and nREM sleep remained significant (p = 0.003).

Periodic breathing. A total of 57 episodes of spontaneous PB were observed in all but one lamb. In 53 episodes of PB, central apneas were present at nadir of PB (Figure 3). In the remaining four episodes, nadir of PB corresponded to hypopneas (1 cycle/2 s) alternating with contiguous breaths. The total duration of PB episodes varied from 58 s to 35 min 12 s (mean = 5 min 11 s ± 3 min 52 s) with 6 to 97 oscillations (hypopnea or apnea) per episode. The percentage of time spent in PB (Table 2) was significantly lower in REM compared with nREM sleep (p = 0.007) but not significantly different between nREM sleep and wakefulness (p = 0.06). During wakefulness, PB was observed typically in the minutes following transition from nREM sleep, when lambs were still drowsy, and disappeared with complete awakening. The percentage of time spent in PB did not change with age, even when controlling for state of consciousness (p = 0.88).

View larger version (52K): [in this window] [in a new window]   Figure 3.   Active glottic closure during periodic breathing in a preterm lamb (postconceptional age = 148 d) during nREM sleep. See Figures 1 and 2 for abbreviations. TA EMG is present throughout all central apneas during periodic breathing. Complete glottic closure during apneas is suggested by absence of cardiac artefact on airflow trace.

Periodic breathing episodes often occurred while lambs were hypoxic, whereas isolated apneas often occurred while lambs were normoxic. Mean Sa_O2 was 89.6 ± 8.8% at the beginning of PB and gradually decreased throughout the episode to reach a mean value of 78.3 ± 11.9%. There was no significant relationship between duration of PB episode and severity of hypoxia (p = 0.54).

Thyroarytenoid Muscle and IPC EMG During Apneas

Central Apneas.

Thyroarytenoid muscle. As a whole, TA EMG was present in 97% of all central apneas. Percentage of apneas with TA EMG and whether or not it was continuous throughout apnea are reported for each state of consciousness in Table 1. Continuous TA EMG throughout apnea was observed in 88.4% of all apneas and 89.5% of central apneas, the percentage of apneas with continuous TA activity being significantly lower however in REM than in nREM (p < 0.0001). No association between age and percentage of apneas with TA EMG was found (p = 0.18), even when controlling for the state of consciousness.

In a few cases, a light decrease in TA EMG amplitude was observed during apnea. When TA EMG was continuous during apnea, the cardiac artefact was never present on the highly amplified nasal airflow trace, suggesting complete upper airway (presumably glottic) closure. On the contrary, in every instance in which TA EMG disappeared during a portion of or throughout apnea, the cardiac artefact appeared on the nasal flow trace (Figure 4). Moreover, in one preterm lamb recorded also using respiratory inductance plethysmograph, observation that the apneic volume was maintained above prior end-expiratory lung volume when continuous TA EMG was present during central apneas (Figure 5) further suggested complete upper airway (presumably glottic) closure.

View larger version (37K): [in this window] [in a new window]   Figure 4.   Active glottic closure during central apnea preceded by a sigh in a nonsedated preterm lamb (postconceptional age = 142 d) during wakefulness. See Figures 1 and 2 for abbreviations. SpO2: transcutaneous oxygen saturation. Following a sigh (high-amplitude diaphragm EMG on the left part of recording), TA EMG is present during the initial portion of central apnea. Initial absence of cardiac artefact on nasal airflow signal (when TA EMG is present), and reappearance of cardiac artefact when TA EMG disappears suggest initial complete glottic closure during apnea

View larger version (36K): [in this window] [in a new window]   Figure 5.   Maintenance of high apneic lung volume by active glottic closure during central apneas during periodic breathing in a nonsedated preterm lamb (postconceptional age = 140 d) during nREM sleep. See Figures 1 and 2 for abbreviations. Active glottic closure (TA EMG and absence of cardiac artefact on airflow signal) throughout recurrent central apneas is responsible for maintenance of high apneic lung volume above prior end-expiratory lung volume.

Whereas TA EMG was most often continuous during apneas not preceded by a triggering event, this was not the case during apneas triggered by an event such as a sigh, a head movement, or a swallow. Thus, most apneas with noncontinuous TA EMG usually followed a sigh (109 of 177), TA EMG being present in the first third of the apnea in 104 cases. Thirty-two of 66 apneas with no TA EMG followed a sigh. Results on TA EMG patterns during apneas preceded by an event, according to state of consciousness, are reported in Table 3.

IPC muscle. IPC muscle EMG was observed during 53% of all apneas, i.e., less frequently than TA EMG. When present, IPC EMG was generally continuous (928 of 1,073 apneas), and always accompanied by continuous TA EMG. The percentage of apneas with continuous IPC EMG was significantly lower in REM than nREM (p < 0.0001). There was no relationship between age and percentage of apneas with continuous IPC EMG (p = 0.10), even when controlling for state of consciousness (p = 0.11).

Mixed apneas. Thyroarytenoid muscle and IPC EMG were recorded in all but three mixed apneas. Amplitude of the continuous TA EMG was constant throughout 37 mixed apneas but decreased in the remaining 28 apneas. A continuous IPC EMG was also observed in 20 mixed apneas.

Periodic breathing. A continuous TA EMG was observed in more than 97% of apneas during PB episodes, regardless of the state of consciousness (Figure 2 and Table 2). Furthermore, expiratory TA EMG with expiratory airflow braking was present during respiratory efforts in all but two PB episodes.

TA Muscle EMG, O₂ Desaturation, and Bradycardia

As a whole, O₂ desaturation and bradycardia were observed in respectively 60.4% (1,261/2,088) and 4.3% (90/2,088) of apneas. The frequencies of these events were correlated with the frequency of apneas longer than 6 s (desaturation: r = 0.69 with p < 0.0001; bradycardia: r = 0.30 with p < 0.0001), as well as the frequency of apneas shorter than 6 s (desaturation: r = 0.68 with p < 0.0001; bradycardia: r = 0.46 with p < 0.0001). Desaturation events seemed to occur more frequently for apneas longer than 6 s, although not statistically significant (Figure 4).

The percentage of apneas with O₂ desaturation was significantly higher in nREM sleep (0.62) compared with REM sleep (0.13, p = 0.0003), but not when compared with wakefulness (0.25, p = 0.09). Moreover, the percentage of apneas with bradycardia was significantly higher in nREM sleep (0.67) compared with wakefulness (0.04, p = 0.01), but not when compared with REM sleep (0.29, p = 0.66). The frequency of continuous TA EMG was positively associated to the frequency of O₂ desaturation (p < 0.0001), even when controlling for age and state of consciousness. Moreover, the frequency of O₂ desaturation was positively associated with age (p = 0.015), even when controlling for state of consciousness (p = 0.03).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The present study brings new knowledge on ventilation and laryngeal dynamics in nonsedated preterm lambs during different states of consciousness, including ventilatory characteristics in preterm lambs born at 130 d of gestation from birth to a postconceptional age equivalent to normal term; and active glottic constriction patterns during ventilatory and apneic periods, including periodic breathing. Of particular interest is the observation that continuous glottic constrictor muscle EMG is present throughout most of the 2,088 spontaneous apneas recorded, both isolated and within periodic breathing epochs.

The Preterm Lamb as a Model of Immature Respiration

Previous studies on spontaneous ventilation in immature newborn mammals are scarce and do not report details on ventilatory pattern throughout states of alertness (14). Despite advanced neurological maturation relative to human newborns, the preterm lamb born at 130 d of gestation appears to be a worthwhile model of immature respiration and neonatal apneas bearing unique similarities to apneas in preterm humans. These include: (1) numerous spontaneous apneas, both isolated and within periodic breathing; the novel observation of prolonged and repeated spontaneous periodic breathing epochs in a newborn mammal is particularly noteworthy; (2) significant desaturation despite mainly short apneas; (3) frequent life-threatening apneas postoperatively; (4) sensitivity to caffeine and doxapram. The preterm lamb should soon prove to be a worthy model to study various issues related to immature control of ventilation, including a better understanding of the determinants of neonatal apneas and various treatment options for apneas of prematurity.

A few attempts to study preterm lambs before 129 d of postconceptional age were not successful because of untractable respiratory distress at birth despite antenatal lung maturation with steroids and TRH. Additional injections of steroids could probably improve the survival of more premature lambs (17). Nevertheless, this would likely lead to intrauterine growth retardation, which is known to have its own consequences on neonatal ventilation (18).

Expiratory TA Muscle EMG

TA muscle expiratory EMG was often observed in nREM sleep and quiet wakefulness in preterm lambs. As reported previously by Mortola in various newborn mammals (19), TA expiratory EMG in preterm lambs appeared to be related to breathing frequency, rapidly disappearing when respiratory rate increased. Furthermore, in agreement with previous data in term lambs (20), TA EMG was sleep state-dependent, being very rare during REM sleep and active wakefulness. Such glottic constrictor expiratory EMG has been previously observed in lambs (20) and dog pups (21). Consequent expiratory airflow braking, also observed in human newborns (22- 24), would prevent gas from flowing out freely from the lungs, hence maintaining an FRC greater than that resulting solely from respiratory system elastic properties in preterm newborns.

Continuous TA Muscle EMG During Spontaneous Apneas

As a whole, continuous TA EMG was observed throughout 88% of more than 2,000 spontaneous apneas, either isolated or during periodic breathing episodes, in nonsedated preterm lambs, irrespective of the state of consciousness.

A few studies in newborn mammals have already reported continuous TA EMG during spontaneous central apneas. In unanesthetized or lightly anesthetized lambs born at term, Harding and coworkers (20) observed continuous TA EMG throughout a few spontaneous central apneas in REM sleep, with TA EMG ceasing when the next inspiratory effort was made. Moreover, an inspiratory breath-holding pattern, which was presumed to correspond to active glottic closure, was recorded in the unanesthetized very immature suckling opossum (25), and in dog pups, kittens, and rat pups (19). However, probably because of technical restrictions due to very small preparations (25) or the small number of spontaneous apneas displayed by animals born at term (19, 20), those observations did not receive much attention at the time.

Though the difference remains unexplained, our frequent observation that TA EMG was present during only the first part of central apneas preceded by a sigh in preterm lambs is in agreement with previous reports. Hence, this has been observed in term lambs during nREM sleep (20) and dog pups in nREM sleep and wakefulness (21).

We observed only a few mixed apneas of short duration. This might be due to the advanced neurological maturation of the preterm lambs and/or to difference in upper airway anatomy compared with premature human infants. Nevertheless, glottic closure was recorded in 96% of these apneas at the moment of inspiratory effort.

Glottic Closure During Periodic Breathing

The present results clearly show that continuous TA EMG was present during PB in nonsedated preterm lambs, whatever the state of consciousness. This confirms previous personal results in induced, posthypoxic PB in term lambs (26). Also, our results are in agreement with an observation in one human infant studied by endoscopy during sleep, in whom glottic closure was observed during recurrent central apneas within an episode of periodic breathing (27).

Active Glottic Closure During Apneas

Several personal observations strongly suggest that TA EMG corresponds to active complete glottic closure during spontaneous apneas. First, TA EMG was consistently accompanied by disappearance of the cardiac artefact on the highly amplified nasal airflow signal during central apneas. In a recent study in anesthetized and paralyzed lambs with endotracheal tube, we have tested the value of the cardiac artefact on highly amplified airflow signal to assess airway closure (unpublished results). Using a similar thermocouple with same amplification as in the present study, we observed that the cardiac artefact remained present as long as the airway diameter was greater than 0.5 mm. Second, preliminary recording in one preterm lamb showed that the apneic lung volume was maintained above preceding end-expiratory lung volume when TA EMG was present during spontaneous apneas (Figure 5). Third, we have shown in a recent series of experiments using endoscopy that complete glottic closure, positive subglottal pressure, and breath-holding at high lung volume consistently paralleled continuous TA EMG throughout artificially induced apneas (unpublished results).

Other results in the literature add to our strong belief that active complete glottic closure can be present during spontaneous neonatal central apneas. In term lambs, positive subglottal pressure (5 to 10 cm H₂O) was recorded simultaneously to TA EMG during a few apneas in REM sleep, attesting to a complete active glottic closure (21). In unanesthetized suckling opossums, inspiratory breath-holding was accompanied by tracheal pressure rises well above atmospheric pressure during apneas, strongly implicating upper airway (presumably laryngeal) closure (25). Furthermore, complete glottic closure was observed during recurrent central apneas within an episode of periodic breathing in one human infant studied by endoscopy during sleep (27).

Previous studies by others have shown that upper airway closure can occur at the pharyngeal level during spontaneous apneas in preterm human infants (28, 29). Our study design did not enable us to conclude if pharyngeal closure (active or passive) occurred during spontaneous apneas in preterm lambs, either simultaneously to glottic closure or alone. Given that the management of apneas should probably be different whether passive pharyngeal closure or active glottic closure is present, future studies will have to delineate which apneas are accompanied by one type of upper airway closure or the other in preterm infants.

As recently underlined by others in discussing results obtained in human preterm infants (11), the presence of upper airway closure during apneas renders labeling of apneas as obstructive or central difficult. In the present study, we have chosen to label all apneas without diaphragm phasic inspiratory EMG as "central," according to classical definition of apnea type. However, it is clear that characterization of apneas should be revised, taking into account the possibility of upper airway closure without respiratory efforts.

Origin of active glottic closure during apneas. According to the widely accepted "three phase" model of respiratory control developed by Richter (30), glottic constrictor muscles are activated during early expiration and inhibited during late expiration. Moreover, a central apnea is generally considered as a prolonged expiration. This suggests that TA EMG should be present only during the first part of apnea. However, this is observed only in apneas following a preapneic event such as a sigh (21, 22). In the vast majority of apneas recorded in preterm lambs, active glottic closure is present throughout apnea. Although the central mechanisms responsible for glottic closure are unknown, phylogenetic and ontogenetic links are readily apparent. Indeed, from an ontogenetic standpoint, apneas of prematurity may represent persistence of a fetal type of "respiration" that alternates prolonged apneic periods with active glottic closure and ventilatory periods (31). From a phylogenetic standpoint, transition from aquatic to terrestrial life became possible for vertebrates when they acquired the ability to breathe air intermittently (lungfish). This type of intermittent ventilation, alternating pulmonary ventilation and prolonged inspiratory breath-holding with active glottic closure, is still observed today in amphibians, reptiles, and many diving mammals, even when they are on land (32). According to Bartlett's suggestion (32), while that primitive breathing pattern may be mainly vestigial in adult (nondiving) mammals, it likely remains important in newborn mammals, including humans. Current knowledge suggests that prolonged periods of apnea, with the exchanger full of gas, are fundamental components of the basic breathing pattern in vertebrates (33).

Potential Physiological Consequences of Active Glottic Closure During Central Apneas in the Neonatal Period

Potential beneficial effects. Previous studies have shown that apneas with severe hypoxemia are characterized by a decrease in lung volume under the previous end-expiratory lung volume (34, 35). Conversely, active glottic closure during central apneas would be beneficial by preventing air from flowing out of the lungs, hence maintaining higher alveolar gas stores for continuing gas exchange during apnea. As we have outlined, arguments drawn from maintenance of high apneic lung volume and positive subglottal pressure when TA EMG is present during induced apneas support this hypothesis. This would be especially favorable in preterm mammals, in which alveolar O₂ stores are low relative to their high metabolic needs. Furthermore, an increase in apneic lung volume during PB would theoretically prevent further ventilatory instability and therefore tend to decrease duration of PB episodes.

Potential deleterious effects. A few observations suggest that active glottic closure during spontaneous neonatal apneas, and aside from the laryngeal chemoreflex, might have adverse effects. Conceivably, if inhibition of glottic constrictor motor neurons is not effective at the end of a central apnea, respiratory efforts will resume against a closed glottis, thus converting the central apnea into a mixed apnea. In keeping with this hypothesis, we have observed a few inspiratory diaphragmatic efforts despite continuation of TA EMG during spontaneous central apneas in fetal lambs in utero (5) and in preterm lambs (present study). Also, obstructive apneas against inspiratory breath-holding have been reported in very immature, unanesthetized suckling opossums (25). Finally, such a mechanism might explain the endoscopic observations of a few mixed apneas against a closed glottis reported by Ruggins and Milner in preterm neonates (6) and in infants with apparent life-threatening events (27). However, the present results show that such a sequence of events occurs rarely in healthy preterm lambs, at least in our experimental setting.

In conclusion, results of the present study strongly suggest that active, complete glottic closure is frequent during spontaneous central apneas, including periodic breathing, in an ovine preterm newborn model. This important novel observation raises numerous questions relative to its physiological and clinical consequences in the neonatal period.