Morphologic Changes in the Upper Airway of Children during Awakening from Propofol Administration

The Research Subjects’ Review Board of the University of Rochester (Rochester, New York) approved this study, and written consent was obtained from parents of each child studied. Children scheduled for elective MRI examination of the head or neck during propofol administration were eligible to participate. Exclusion criteria included a history of sleep apnea, nighttime snoring, tonsillar hypertrophy, presence of obesity, or any type of upper airway or external cranial pathology (e.g. , macrocephaly). Premedication was not administered. A standard mask anesthetic technique was accomplished using sevoflurane and nitrous oxide. After consciousness was lost, an intravenous catheter was placed, sevoflurane and nitrous oxide were discontinued, and a continuous infusion of propofol, 200 μg · kg⁻¹· min⁻¹, was begun. Spontaneous ventilation was maintained throughout the procedure. A nasal cannula provided oxygen, usually at 2 l/min. Monitors included electrocardiography, pulse oximetry, capnography (via  the nasal cannula), and intermittent automated blood pressure determinations (Medrad, Inc., Indianola, PA). A best attempt was made by the anesthesiologist to position the child's head in the neutral position (without flexion or extension), usually with a folded sheet as a roll beneath the neck or shoulders. No devices that assist with patency of the airway were used, and no attempt was made to open or close the mouth. At the completion of the indicated scans (> 45 min after discontinuation of sevoflurane), additional sagittal and axial views of the upper airway that extended down to the level of the vocal cords were obtained (this sequence of scans took 112 s to complete). This represented the configuration of the upper airway during sedation. No attempt was made to obtain images at a certain point in the respiratory cycle. After these scans were completed, the propofol was discontinued, and exactly 5 min later, the same scans of the upper airway were repeated as before. These represented the awakening views. If the child began to move during these last scans, the study was terminated, and the child was allowed to awaken. All scans were stored in the MRI computer until studies had been performed in 21 subjects.

Magnetic resonance imaging was performed on a 1.5-T unit with a quadrature head coil (GE Medical Systems, Milwaukee, WI). T1 spin-echo images were obtained with 3-mm slices at 0.5-mm intervals.

Initially, a coronal view of the head was used to identify the most appropriate medial sagittal slice through the upper airway. Using this medial sagittal image, we identified the levels of the most posterior edges of the pharynx at the levels of the soft palate and tongue in the anterior–posterior (A-P) direction. We then used the axial slices that were closest to these levels to measure in millimeters the A-P and transverse diameters and the cross-sectional areas (figs. 1–3). Linear measurements of diameters were made directly using a GE Horizon 5X scanner, version 5.8 software (General Electric Co., Milwaukee, WI). Cross-sectional areas were directly calculated by tracing freehand the edges of the airway. The measurement software reported distances in millimeters to the nearest whole integer. Measurements were not performed at the level of the epiglottis because it is situated within the laryngeal pharynx and is not bounded laterally by any structure.

After all data were collected, each individual scan was assigned a number, and these numbers were then randomized to determine the order of measurement. Measurements were made by at least two of the authors who were blinded to the state of unconsciousness. When there was disagreement about the value of the distance measured, a final value was determined by consensus, by assistance from a third author, or both.

Statistical calculations were performed by SigmaStat for Windows (Jandel, San Rafael, CA). Comparisons between sedated and awakening states were assessed using two-tailed paired t  tests for normally distributed data and the Wilcoxon Signed Rank test for nonnormally distributed data. A P  value less than 0.05 indicated statistical significance.

Twenty-one children initially enrolled, and five were excluded because of movement artifact (three) or early awakening (two). Therefore, data are presented for 16 children. There were nine boys and seven girls, whose age ranged from 10 months to 7 yr (mean ± SD = 3.2 ± 2.0) and whose weight ranged from 9 to 23 kg (mean ± SD = 14.6 ± 4.9).

In the sedated state, we were able to measure the cross-sectional area of the pharynx at the level of the soft palate and tongue in 14 children (2 were technically poor images). The cross-sectional area was smallest at the level of the soft palate in 11 children and smallest at the level of the tongue in 3 children. At the level of the soft palate, the transverse diameter was most narrow in 11 children, the A-P diameter was most narrow in 1 child, and they were equal in 2 children. At the level of the tongue, all 16 children were measurable. The transverse diameter was most narrow in 9 children, the A-P diameter was most narrow in 5 children, and they were equal in 2 children.

During awakening, we were able to measure the cross-sectional area of the pharynx at the level of the soft palate and tongue in 14 children (2 were technically poor images). The cross-sectional area was smallest at the level of the soft palate in 12 children and smallest at the level of the tongue in 2 children. At the level of the soft palate, the transverse diameter was most narrow in none of the children, the A-P diameter was most narrow in 13 children, and they were equal in 1 child. At the level of the tongue, all 16 children were measurable. The transverse diameter was most narrow in 4 children, and the A-P diameter was most narrow in 12 children.

Comparison of the dimensions of the pharynx at the levels of the soft palate and tongue revealed significant differences between sedated and awakening states (fig. 4). During awakening, the A-P diameter of the pharynx at the level of the soft palate decreased in 12 children, increased in 1 child, and remained the same in 1 child. (P < 0.001). The transverse diameter increased in 11 children, decreased in 1 child, and remained the same in 2 children (P = 0.001). The cross-sectional area at the level of the soft palate increased in 4 children, decreased in 8 children, and stayed the same in 2 children (P = 0.5).

During awakening, the A-P diameter of the pharynx at the level of the tongue decreased in 11 children, increased in 4 children, and remained the same in 1 child. (P = 0.01). The transverse diameter increased in 11 children and decreased in 5 children (P = 0.07). The cross-sectional area at the level of the tongue increased in 7 children, decreased in 7 children, and stayed the same in 2 children (P = 0.9).

During either phase of the study (sedated or awakening), no child demonstrated clinically significant upper airway obstruction as manifested by a decreased oxyhemoglobin saturation or decrement in capnometric values. Clinical indicators of upper airway obstruction (e.g. , stridor, use of accessory chest wall muscles, and so forth) were not assessed while the children were inside the scanner.

The major finding of this study is that the dimensions of the upper airway change shape significantly on awakening from propofol administration. When sedated, the upper airway is oblong shaped with the A-P diameter larger than the transverse diameter. On awakening, the shape of their upper airway in most children we studied changed such that the transverse diameter was larger. A comparison of the cross-sectional area between sedated and awakening states revealed essentially no change. Unfortunately, a control group consisting of conscious children with which to compare these findings was not included because of the near impossibility of scanning a child without some form of sedation.

A secondary finding was that while sedated, most children had the most narrow site in their pharynx at the level of the soft palate. In the past, upper airway obstruction during general anesthesia was thought to occur at the level of the base of the tongue, presumably because of anesthetic-induced genioglossus muscle relaxation. 4However, subsequent studies in both adults and children showed the major site of airway narrowing to be at the level of the soft palate or epiglottis. 1,5–8None of these previous studies investigated the anatomic relations of the structures of the upper airway during awakening.

The mechanism of these changes during awakening remains speculative. It is generally accepted that patency and stability of the upper airway are dependent on contraction of the muscles that comprise the nonbony portions of the pharyngeal wall. The start of the contraction of these pharyngeal dilators slightly precedes the contraction of the diaphragm to counteract the negative pressure generated during spontaneous respiration. 9Furthermore, there is evidence that the caliber of the upper airway increases during inspiration because of thoracic inspiratory activity providing caudal traction on the upper airway. 10Although the effects of general anesthesia reduce overall ventilatory drive and airway reflex activity, pharyngeal muscle activity may be more susceptible than that of the diaphragm. 3Administration of halothane, thiopental, and diazepam results in differential effects on strength of the hypoglossal and phrenic nerves, but ketamine does not share this differential property. 3Propofol has been shown to decrease diaphragmatic contractility in an animal model, 11but its differential effects on upper airway musculature and diaphragmatic strength have not been investigated.

Another explanation is offered by a previous study that used radiography to investigate the effect of a sleep dose of thiopental on the structures of the upper airway during spontaneous ventilation. 12These authors found anterior displacement of the epiglottis relative to the cervical spine and widening of the glottic aperture with the onset of general anesthesia. They suggested that the tonic activity that “folds” the larynx during consciousness is a protective mechanism and is lost with induction of general anesthesia.

A limitation of this study relates to the difficulty of studying the subjects in a consistent state of awakening. Because the pharmacokinetic and pharmacodynamic properties of propofol vary in different individuals, it would be difficult to precisely target a specific state of awakening. Our extensive clinical experience using this technique indicated that children almost always awaken 6–8 min after discontinuation of propofol when infused at this particular dose for the duration of a typical head scan (45–60 min). Therefore, we chose to obtain the awakening scan exactly 5 min after discontinuation of propofol in the hope of capturing the subjects at a stage of being nearly awake. There are two disadvantages of this technique. The first is that some children may awaken fully and begin to move before completion of the study (two enrolled subjects were excluded for this reason). Second, and more important, is the possibility of studying a subject in an advanced state of awakening whereby a higher level of cortical function is influencing the upper airway musculature. In this state, it is conceivable that these muscles might be in a state that renders them less prone to cause airway obstruction. This latter explanation may account for the observed variation between subjects we studied. The majority of the children who completed the study were able to lie motionless during the loud noise of the scanner but then awoke immediately when they were brought out of the scanner and exposed to the physical stimulation of monitor removal. An alternative explanation for our results is that as the children began to awaken, swallowing could have occurred, which caused the narrowing appearance on MRI.

Another limitation of our study is that all children were studied with their necks in a neutral position. Therefore, our results would not necessarily apply to children whose necks are flexed or extended or who are receiving forward mandibular displacement. These maneuvers have been shown to alter the shapes and positions of the structures of the upper airway. 1Likewise, the degree of mouth opening can alter the relations of the anatomic structures of the upper airway. We made no attempt to standardize mouth opening in this study. However, our clinical experience has been that the majority of children who receive propofol sedation for MRI maintain their jaw in the neutral position, somewhat midway between opened and closed.

In summary, we demonstrated morphologic changes in the shape of the upper airway of children who were awakening from propofol sedation. These findings provide further evidence of the complexity of the effects of anesthetics on upper airway musculature and suggest that propofol, like other anesthetics and sedatives, may also be associated with a differential effect not only between the diaphragm and upper airway, but between separate pharyngeal dilator muscles.

The authors thank Traci Salter, R.T.R. (Department of Radiology, Strong Memorial Hospital, Rochester, New York), and the team of magnetic resonance imaging technicians at Strong Memorial Hospital for making this study possible.