Effects of Pain and Audiovisual Stimulation on the Opioid-induced Depression of the Hypoxic Ventilatory Response

The study was completed at the University of Rochester Medical Center (Rochester, New York) and approved by its human subjects protection committee. Written, informed consent was obtained from all subjects. Subjects were asked to fast from midnight before the experiment and to refrain from alcohol- and caffeine-containing beverages for 24 h before the experiments. Most of the subjects were experienced from other experiments in our laboratory and were familiar with the laboratory environment.

At the time of arrival at the laboratory, all subjects had an intravenous catheter inserted for administration of remifentanil. A pulse oximeter was placed on a finger of the nondominant hand because the dominant hand was used for the game controller. A three-lead electrocardiogram was placed on the chest wall for heart rate and rhythm monitoring, and blood pressure was measured using an automated cuff. The subjects were then allowed to relax before the first experimental intervention.

The techniques of the respiratory measurements have been described in detail previously.14Briefly, subjects were semirecumbent, with electrocardiography and pulse oximetry (Siemens Medical Electronics, Danvers, MA) monitored continuously for subject safety. Blood pressure was measured on arrival, before and after each run, every 5 min during the remifentanil infusion (but not during measurement of the AHR), and before discharge. Subjects breathed from a gas-mixing chamber via  a facemask (Vital Signs, Totawa, NJ). Inhaled and exhaled volumes were measured with a bidirectional impeller flow meter (VMM 110; Sensor Medics, Laguna Hills, CA). Airway gases were sampled continuously by a mass spectrometer (MGA 110; Perkin-Elmer, Pomona, CA). Computer-driven (Intel PC class) high-flow (total flow of 60 l/min) stepper-motor valves provided nitrogen, carbon dioxide, and oxygen to the gas-mixing chamber at desired concentrations on a breath-by-breath basis. Ventilation (V^E, l/min), tidal volume (V^T, liters), breathing frequency (f, breaths/min), end-tidal gas concentrations (end-tidal pressure of oxygen [Peto²] and end-tidal pressure of carbon dioxide [Petco²], mmHg), and hemoglobin oxygen saturation by pulse oximetry (Spo², %) were determined and collected using the TIDAL software package.∥15 

To study the ventilatory response to isocapnic hypoxia, a computer-driven dynamic end-tidal forcing technique was used. With this technique, Peto²and Petco²are dynamically forced to follow a prescribed pattern in time by manipulation of the inspired gas concentrations independent of the ventilatory response. In this study, step transitions in Peto²were used with a background of constant Petco²that was individually determined for each subject. The 10-min breathing period started with 5 min of normoxia (Peto²= 100 mmHg) and then a rapid decrease to Peto²of 45 mmHg within two to three breaths; it was then sustained for 5 min.

The hypoxic response was repeated for each subject under six conditions (fig. 1): low wakefulness (LW), high wakefulness (HW), and low wakefulness with pain (LWP), with and without (control) a continuous infusion of remifentanil.

The LW experiments were performed under the conditions of “resting wakefulness,” which is similar to that described by others as the subject having closed eyes and headphones on to prevent auditory stimulation.4,6Electroencephalographic monitoring was not applied, but subjects were asked to remain awake and were queried if they fell asleep at the end of the run. The laboratory was kept quiet, and any movement by laboratory personnel was out of sight of the subjects.

In the HW state, AVS was provided by having the subjects play a computer game entitled “You Don't Know Jack—The Ride” (Berkeley Systems, Jellyvision Inc., Chicago, IL, 1998) on a laptop computer positioned directly in front of them. The audio portion of the game was played through external sound-deadening headsets (Realistic NOVA 40; RadioShack, Fort Worth, TX) that were worn for all experiments (no sound for the pain and rest experiments). The subjects answered the trivia questions posed by the game by pressing a key on a game control pad (minimal movement required). Computer games have been used previously as a means to increase the state of wakefulness.4,16 

For LWP, painful stimulation by a thermode (PPS-3; Precision Pain Source, Cygnus, Paterson, NJ) was added to the LW state. The skin on the ventral surface of the arm, contralateral to the arm with the intravenous catheter, was preheated to a temperature of 40°C for 5 min and then sensitized with 0.075% capsaicin cream under an occlusive dressing for at least 30 min.17,18The subjects reported their pain scores via  an ordinal score that ranged from 0 (no pain) to 10 (intolerable pain). The pain score was the average of the two scores reported at the beginning and end of the experimental run. Subjects were instructed to notify personnel if their pain score was minimal (< 3) during the experimental run by raising a hand. In those cases (n = 2), the run was discontinued and repeated using a new thermal stimulation site on the forearm. During the remifentanil infusion, the temperature of the thermode was increased to maintain the same pain score, but for safety reasons, the maximum temperature was limited to 47°C.

The remifentanil was administered via  a syringe infusion pump (Baxter Infusion Pump, model AS50; Baxter Healthcare Corp., Deerfield, IL). The sequence of the runs was randomized by the Latin square method,19with the limitation that the control and remifentanil experiments were always alternated to avoid a prolonged infusion and possible tolerance to the opioid.20To reach steady state, at least 30 min was allowed after turning off the remifentanil before a control experiment, and the remifentanil infusion was started at least 15 min before a hypoxic response.21 

All subjects participated in preliminary experiments to determine their individual remifentanil dose and appropriate carbon dioxide level for the subsequent hypoxic response. These experiments were either on a different day (n = 5) or earlier in the same day (n = 6) as the main experiment. The selected dose of remifentanil for these experiments was a dose of remifentanil that required a significant increase in the thermode temperature to maintain the same pain score without exceeding the maximum of 47°C, and with an increase in Petco²of approximately 5 mmHg. If the Petco²increased too much with the remifentanil, then the hypoxic response under control conditions with the same Petco²may not have been tolerated well by the subjects. In these preliminary experiments, we used increasing doses of remifentanil starting at 0.025 μg · kg⁻¹· min⁻¹and going to an expected maximum of 0.1 μg · kg⁻¹· min⁻¹and measured the pain score, ventilation, and Petco². A dose of remifentanil that adequately blunted the pain while increasing the Petco²by approximately 5 mmHg was determined in all subjects. For each subject, this dose of remifentanil was subsequently used for all of the experiments. The isocapnic AHR was performed at a Petco²level 2–3 mmHg higher than that which was individually determined in these preliminary experiments.

The respiratory measurements were averaged over the 2 min before the hypoxic step and for 2 min between minutes 2 and 4 of the hypoxic period. The saturation was calculated with use of the measured Peto²using the equation suggested by Severinghaus.22The acute hypoxic response was calculated as the change in ventilation divided by the change in saturation.

Data are reported as mean ± sd. Statistical analysis was performed using the STATA software package (Stata Corp., College Station, TX). Initial analysis of variance was performed on the AHR and the other ventilatory variables using the following factors: sex, subject nested within sex, test condition (LW, LWP, or HW), drug (control or remifentanil), test condition and sex interaction; and drug and sex interaction. Subsequent analysis of variance dropped nonsignificant terms. When a significant test effect was found, a post hoc  Wald test was used to isolate the differences for pairwise comparisons. Overall significance for multiple comparisons was set at 0.05 using the false discovery rate procedure.23One-way analysis of variance was used to test for a drug effect within a test condition.

Seventeen paid volunteers (nine men) who had an American Society of Anesthesiologists physical status of I, were aged 18–40 yr, were nonsmokers, and took no medications other than birth control pills were enrolled. Eleven subjects completed the protocol (six male and five female subjects). Six subjects did not complete the study: Two subjects dropped out because of nausea and diaphoresis with the remifentanil, one subject's experiment was aborted because of his vomiting, one subject elected not to return for unknown reasons, one subject experienced syncope during placement of the intravenous cannula, and the computer malfunctioned during the remaining subject's experiment. The average age was 21.2 ± 2.1 yr, and the average body mass index was 24.0 ± 2.7 kg/m². During the experiments with pain, the temperature was successfully adjusted to maintain similar pain scores, but in two subjects, it required using a new forearm site because of inability to stimulate these subjects to a high enough pain score. In two subjects, a blister subsequently developed at the thermode site, but this resolved without consequence. No subject reported any painful sensation from the capsaicin cream before the application of the thermode.

The average dose of remifentanil, as determined by the preliminary experiments, was 0.035 ± 0.012 μg · kg⁻¹· min⁻¹, and the dose was not significantly different between the male and female subjects. The remifentanil infusion required a significantly higher temperature (46.1 ± 1.1°vs.  44.0 ± 2.0°C; P < 0.05), whereas the pain scores were not significantly different (3.8 ± 0.7 vs.  4.2 ± 0.8 for remifentanil and control, respectively). All subjects reported remaining awake during the resting runs. During the remifentanil infusion, the subjects consistently progressed through the computer game and never needed reminding to play the game.

The end-tidal carbon dioxide and saturation were well controlled across states and drug (table 1). The sd in Petco²shown in table 1reflects primarily the variation across subjects, each having their own carbon dioxide set point. There were some small differences that achieved statistical significance because of the consistency of the gas control system we used. However, these differences were generally less than 1 mmHg in carbon dioxide or 1% saturation and did not significantly affect the results.

Figure 1shows the superimposed breath-to-breath results of the six hypoxic steps in one subject. This subject had a pronounced increase in the ventilatory response to hypoxia during HW but a decrease during LWP. The hypoxic response was clearly depressed by the remifentanil infusion in all states.

Overall analysis of variance on the AHR (fig. 2and table 1) showed significant state and drug effects but no sex, state and sex interaction, or drug and sex interaction. Therefore, subsequent analysis did not include sex or its interactions, but the small number of subjects limits our ability to detect a sex effect if actually present. The drug and state interaction was not significant (P = 0.06), and the final statistical analysis was done within each treatment group (remifentanil or control). The AHR during HW was significantly increased over low wakefulness with and without pain. Although the presence of pain caused a decrease in the AHR over LW without pain, this was not statistically significant. The remifentanil infusion reduced the AHR in all three states, and the HW state was still significantly different from the others (table 1). The increase in AHR (either directly or as a percentage of the LW value) with HW from LW was not significantly different between control and remifentanil.

The results for normoxic ventilation were similar to the result for AHR (table 1). HW significantly increased both normoxic and hypoxic ventilation for both control and remifentanil. Although pain slightly increased normoxic ventilation during control experiments, this was not statistically significant. Looking at the pattern of breathing, remifentanil significantly decreased the tidal volume, with a decrease in breathing frequency that usually did not reach statistical significance. The effect of HW was seen in both tidal volume and breathing frequency during normoxia and primarily in breathing frequency during the remifentanil infusion. Interestingly, comparing across drug conditions, the breathing frequency during remifentanil and HW was significantly higher than control LW in both normoxia and hypoxia.

The heart rate response followed a similar pattern (table 1). Remifentanil reduced the heart rate under all conditions, although not significantly during LW in normoxia and HW in hypoxia. Similar to ventilation, both with and without remifentanil, HW caused an increase in heart rate over LW, whereas LWP did not.

In a single study, we compared the effects of pain and audiovisual stimulation on the opioid-depressed acute isocapnic hypoxic ventilatory drive. The most interesting finding in this study, and contrary to our hypothesis, is that although HW augmented the AHR in the presence of remifentanil, the reversal of the depressed AHR was not complete and was less than the increase caused by HW without remifentanil (fig. 2). Pain had little effect on the AHR, even in the presence of remifentanil. However, HW during the remifentanil infusion did restore normoxic ventilation to that found during LW without drug.

There are many factors in the design of this protocol that could greatly influence the results and make comparisons with previous studies difficult. The study design is complex, and there are potential confounding effects in the sequence of runs for both the remifentanil infusion and the stimulation.

A randomization sequence was selected that should have minimized these effects in the final data analysis. In particular, remifentanil experiments were alternated with control (no drug) runs to avoid prolonged infusions that might cause acute tolerance, whereas the sequence of HW, LW, and LWP was randomized. The techniques we used to assess the hypoxic ventilatory response are standard24,25and frequently used in our laboratory.14The resting values for ventilation and the AHR, with and without remifentanil, are comparable to values in the literature. Specific issues regarding the selection of the drug and stimuli will be addressed in the next sections.

Remifentanil has been reported to cause persistent hyperalgesia and tolerance under certain conditions. The remifentanil dose used was below the reported infusion rate that causes intense opioid receptor activation and possible persistent increase in pain sensitivity, or hyperalgesia.26In a study of 20 male volunteers receiving a 3-h infusion of remifentanil at 0.08 μg · kg⁻¹· min⁻¹(above our average dose), there was no development of acute opioid tolerance.20Therefore, it is unlikely that the dose and sequence of remifentanil infusions distorted subsequent experiments.

Although it may be assumed that the ascending pain pathways can influence both the metabolic and the behavioral control of breathing, there is little information regarding differences in type of painful stimulation. Many experimental pain models have been used in respiratory control studies, including electrical stimulation,7,10,13pressure,27and limb ischemia7,8,28,29in humans and thermal stimulation in animals.30Electrical pain may stimulate both cutaneous and muscle receptors, but remifentanil causes more of an increase in muscle versus  cutaneous pain thresholds.31Interestingly, activation of muscle versus  cutaneous pain receptors may have greater ventilatory effects.7Cutaneous pain, whether electrical or thermal,32may not be a reliable indicator of clinical and acute postoperative pain and is different from simulated visceral pain as well.33Because there is no consensus about the best model, the capsaicin-sensitized skin thermal model was selected because it has been well described as being sensitive to remifentanil analgesia.18 

The choice of a painful thermal stimulus may not have permitted a noxious enough stimulus to elicit the augmentation of ventilation. However, the reported pain scores were in the moderate range and similar to those reported for other types of painful stimulation.10The thermal stimulation was increased during the remifentanil infusion to maintain the same reported pain level. This was done to maximize the effect of pain on breathing as has been previously shown by Waldrop et al.  30; presumably, if the thermal stimulation was not adjusted, the reported pain scores would have been lower during the remifentanil infusion, and even less effect of pain on breathing would have been observed.

Although the type of painful stimulus was different than that of Sarton et al. ,10the results are similar in that the painful stimulus did not significantly alter either normoxic or hypoxic ventilation. This differs from the results of Pandit et al. ,13who found a small stimulation of both normoxic ventilation and the AHR with their painful stimulus. Both Sarton et al.  and Pandit et al.  used electrical stimulation to arrive at similar pain intensities.

The mechanism of action of pain on breathing is still enigmatic. The interaction of pain and breathing seems to be mediated primarily via  the effect of pain on normoxic ventilation and not on the chemoreflexes.7,34–36Pain has been noted to elicit a ventilatory response in the absence of a cardiovascular or somatic response, even while under 1 minimum alveolar concentration (MAC) of inhalational anesthesia, clearly indicating an effect independent of any volitional control.9Interestingly, Lam et al.  36found that pain in the form of surgical stimulation increased baseline ventilation but did not restore the AHR in the setting of 1 MAC enflurane anesthesia. However, we have found that even with volitional control still intact (LWP), the AHR remained depressed by remifentanil.

One could postulate that pain would be expected to more directly counteract the depressive effect of the opioid remifentanil on the AHR. Similar to previous studies, though, our painful stimulation did not affect the chemoreflex drive to breathe.

There is also no consensus on the best experimental model to assess the influence of the arousal state on ventilatory control. Passive forms of cortical activity (mental arithmetic or watching and listening to various videos) may not provide the same level of wakefulness, and it is difficult to assess in an individual subject. In this study, we used an interactive computer game to obtain our central nervous system HW state.4The current choice of audiovisual stimulation is different from what has been commonly used when studying the effects of inhalational anesthetics, because our stimulus required a reaction from the subjects, and its effect on hypoxic breathing may be different as well. We did not define the wakefulness state by an observer's scale or by the electroencephalogram or Bispectral Index, as has been done previously.10Instead, we sought to maximize the wakefulness state as has been described.4,16The subjects in our study were able to play the computer game during the remifentanil infusion without any need for reminders to continue to progress.

Similar to previous studies,10,25we found a significant increase in normoxic ventilation during HW. We formerly reported a significant augmentation of the AHR during HW compared with low wakefulness with and without pain, in a larger cohort of subjects, without infusion of drug.37 

Although ventilation during HW and hypoxia was generally increased over LW and LWP, the subjects subjectively reported less discomfort from the increased ventilation during HW than they did during the other states. This was noted especially when the drug was not being infused. Interestingly, pain has been found to intensify the sensation of dyspnea.28 

The effects of inhalational anesthetics on the hypoxic ventilatory response and the effects of AVS and pain on this interaction have been previously extensively studied.10–12,38A specific reversal effect of AVS on the depression of the AHR has been found for isoflurane11but not for halothane.13In the latter study, though, the subjects did perform the exercise of holding an arm raised, which may have affected their experimental outcome. Pandit12has argued that there may be true differences between the effects on the AHR of the various inhalational agents and that the reported differences cannot be completely explained by the various experimental techniques used. However, even small differences in the study protocols may have greatly affected study outcomes.

In any case, there are obvious differences in mechanisms between the inhalational anesthetics and opioids, although both can achieve a similar reduction in the AHR. One relevant difference is the action of the inhalational anesthetics on the carotid bodies, whereas the depression of the AHR by opioids occurs by a central action because spinal opioids have been shown to depress the AHR.27 

Another difference between the drugs may be the level of sedation produced. Although our study can be criticized for not using the Bispectral Index® monitor (Aspect Medical Systems, Newton, MA) as a quantitative assessment,39the Bispectral Index® may not be a sensitive monitor for opioid sedation levels.40Our subjects were consistently able to maintain wakefulness while playing the computer game and did not report falling asleep during the LW or LWP experiments when they had their eyes closed.

This confirmation of what has been already found in the setting of inhalational anesthetics is even more clinically significant with opioids because the effects of narcosis on patients is longer lasting and a more constant threat than the effects of inhalational anesthetics.

In this study of the remifentanil-induced depression of the isocapnic AHR, an HW state (simulated by playing a computer game) did not restore the AHR to baseline, although it did augment it more than painful thermal stimulation. Clinically, these results could indicate that a specific audiovisual stimulation requiring a volitional patient reaction may be more effective than pain in restoring adequate ventilation in responsive narcotized patients.

In light of the conflicting data resulting from studies of inhalational anesthetic influence on the AHR, the opioid interaction, and the effects of different arousal states, it is clear that future studies are needed to clarify the mechanisms involved.

The authors thank James Robotham, M.D., F.R.C.A. (Chairman, Department of Anesthesiology, University of Rochester, Rochester, New York), for his thoughtful review and discussions during the preparation of this article.