The Absence of Acute Tolerance During Remifentanil Infusion in Volunteers

Abstract

The development of acute opioid tolerance in humans remains controversial. We tested the hypothesis that continuous remifentanil infusion leads to rapid development of opioid tolerance. Twenty healthy male volunteers were enrolled onto a randomized, placebo-controlled, double-blinded, cross-over design study to receive a 3 h continuous infusion of remifentanil (0.08 μg · kg⁻¹ · min⁻¹) or saline. Test procedures included determination of pain perception thresholds and pain tolerance thresholds to heat and cold and neuroselective sine wave constant current at 5 Hz and 250 Hz. Test procedures were performed at baseline and then repeated at 25, 55, 85, 115, and 160 min (heat/cold) and at 35, 65, 95, 125, and 170 min (electrical current) during infusion. No significant decrease of the pain threshold devolutions between 55 and 180 min after the start of infusion of remifentanil could be detected. In conclusion, no development of acute opioid tolerance was observed during constant remifentanil infusion of 3 h in volunteers.

The development of acute opioid tolerance has been shown in animals ⁽¹⁾. Recently, the development of acute opioid tolerance has been suggested in a nonblinded study of healthy volunteers and in patients after opioid-based anesthesia ^(2,3). However, the existence of acute opioid tolerance remains controversial ⁽⁴⁾.

The study of opioid tolerance in humans requires an experimental method that provides a reliable detection of the analgesic effect of opioids. Several approaches have been used to test the analgesic effect of opioids and other drugs in humans, but they have been unsatisfactory because of the high variability of subjective pain ratings ^(5,6). Another alternative for determining the effect of analgesics is quantitative sensory testing (QST). QST has the particular advantage of being a functional test providing a quantitative pain stimulus ^(7,8). The repeatability of the visual analog scale demonstrated poor precision in a setting of human experimental heat pain compared with QST ⁽⁹⁾. QST also provides a reliable assessment of changes in pain thresholds. We recently demonstrated, by means of QST, that remifentanil results in a dose-dependent increase in heat pain thresholds in volunteers ⁽¹⁰⁾.

The purpose of this randomized, placebo-controlled, double-blinded, cross-over study was to test the hypothesis that remifentanil leads to rapid development of opioid tolerance in humans. The occurrence of opioid tolerance was assessed during a constant remifentanil infusion by a decrease of pain perception threshold (PPT), pain tolerance threshold (PTT), or both using thermal (heat and cold) and constant current sine wave electrical stimuli.

Materials and Methods

The study was approved by the Vienna University IRB. Written informed consent was obtained from 20 paid volunteers. Inclusion criteria were healthy ASA physical status I males, aged 19–40 yr, with a body mass index within the 15th and 85th percentile. All subjects underwent a medical interview. Exclusion criteria were any current acute or chronic pain conditions, use of analgesics within 1 week, and a history of drug abuse. Subjects were not allowed any oral intake for 6 h before the drug administration.

Subjects were randomly assigned by computer to two groups receiving consecutively either remifentanil (Group 1) or saline (Group 2) infusions in a cross-over fashion. The order of applications was randomized between the groups. Each subject was studied in two sessions at an interval of at least 5 days.

To avoid bias, volunteers were blinded to the aim of the study. Study sessions were performed in a quiet, unstressful environment at the same air-conditioned location and always at the same time in the afternoon. The same trained observer, blinded to the study medication, supervised all tests.

From the beginning of the study, subjects were continuously monitored for heart rate, respiratory rate, oxygen saturation, and noninvasive arterial blood pressure (right arm). During the administration of the study drugs, a sedation score (0 = awake, 1 = tired, 2 = asleep but arousable, and 3 = nonarousable) was assessed every 10 min. All side effects were noted.

An IV catheter (20-gauge) was inserted in the left cubital vein and connected to a glucose 5% infusion. Before each session, an impartial nurse prepared an indistinguishable infusion syringe containing remifentanil (Ultiva^®, GlaxoWellcome, Vienna, Austria; 20 μg/mL in saline) or saline. The syringes were attached to a continuous syringe pump (Perfusor^® fm, B Braun, Germany) and piggybacked into the glucose 5% infusion. A continuous infusion of either 0.08 μg · kg⁻¹ · min⁻¹ of remifentanil or saline was administered for 180 min.

During each session, oxygen (2 L/min) was applied via a nasal cannula. Infusion was discontinued if: (a) the respiratory rate decreased to <7 breaths/min, (b) peripheral oxygen saturation decreased to <85%, (c) the heart rate decreased to <40 bpm, (d) the mean arterial blood pressure decreased to <60 mm Hg, or (e) excessive sedation occurred preventing adequate handling of the TSA-2001 (Medoc, Ramat Yishai, Israel) or Neurometer^® CPT/C (Neutron Inc, Baltimore, MD) devices (sedation score greater than or equal to two). The subject could also discontinue the infusion voluntarily at any time during the study for any reason.

PPTs and PTTs were assessed by both thermal and constant current sine wave self-administered quantitative sensory testing. Thermal sensory testing was performed using a commercially available thermal sensory testing device (TSA-2001). The Peltier thermode, sized 18 × 18 mm, was attached to the volar aspect of the left forearm. Skin adaptation temperature was 32°C, and rate of temperature change was 0.8°C/s with a return rate of 4°C/s. Stimulator temperature range was 32°C–1°C and 32°C–53°C, respectively. Thresholds were measured through the method of limits as described previously ⁽¹¹⁾. Subjects were initially trained in a standardized manner to perceive the thresholds. Subjects were instructed to stop the decrease or increase of temperature at the first perception of unpleasant cold (cold PPT) or heat (heat PPT). This test was repeated three times and averaged. There was a 30-s interstimulus rest period between each PPT determination. Next the volunteers were instructed to stop the decrease or increase of temperature when the stimulus became intolerable (cold PTT or heat PTT). This test was repeated twice and averaged. There was a 60-s interstimulus rest period between each PTT determination.

Electric current sensory testing was done using the commercially available Neurometer^® CPT/C automated electrodiagnostic device. A pair of 1-cm diameter gold electrodes separated by a 1.7-cm Mylar spreader were coated with a thin layer of chloride-free electroconductive gel and then taped to the distal phalange of the left index finger. Thresholds to two different frequencies of a constant current sine wave stimulus were tested (250 Hz and 5 Hz). Pain threshold measures were performed using a standardized automated double-blinded methodology (Neurometer CPT/C Operating Manual, Neurotron Inc, 1999). The stimulus was presented in an ascending staircase fashion from zero to a maximum of 9.99 mA. The pain thresholds were performed by the subject pressing and holding the “Test Cycle” button on the remote box component of the device. The subject released the button to automatically discontinue the stimulus when the stimulus first became perceived as painful (electrical PPT). This test was repeated three times and averaged. There was a 30-s interstimulus rest period between each PPT determination. Next the subjects were instructed to discontinue the increase in stimulus intensity when the stimulus became intolerable (electrical PTT). This test was repeated twice and averaged. There was a 60-s interstimulus rest period between each PTT determination. The stimulation would automatically stop if the maximum output intensity (9.99 mA) was reached. The duration of each step was a function of the stimulus frequency (2.16 s at 250 Hz [20 steps] and 2.52 s at 5 Hz [29 steps]). The reliability of pain threshold measures using the Neurometer^® CPT/C device has been established in several recent studies ^1,2(12–15). Measurements were performed at baseline before drug infusion and then repeated at 25, 55, 85, 115, and 160 min for the thermal and 35, 65, 95, 125, and 170 min after the electrical current threshold determinations, respectively.

Before starting the study, a statistical a priori planning was performed. Based on a previous study ⁽³⁾, the intention was to detect a decrease of the maximal pain threshold values of 50%, an effect level of 1.0, with an α-error of 5%. Using a two-sided paired Student’s t-test, the study population was calculated to be 16 participants to reach a minimum power of 80% (nQuery software for Windows 95, Statistical Solutions, Boston, MA). Also, based on our own experience using this experimental setting, we expected a dropout rate of 20%. Therefore, the initial total number of volunteers was set to 20 persons.

With the constant infusion of remifentanil, 99% of the steady-state concentration is reached after at least 55 min and is stable thereafter ⁽¹⁶⁾. Therefore, for all pain thresholds, the maximal value was expected to be reached after 55 min by all volunteers. The development of opioid tolerance was defined as a decrease of the maximal pain thresholds at time point two and should be observed at least at time points three, four, and five (85, 115, and 160 min and 65, 95, and 170 min, respectively). Assuming a constant decrease over time for each patient, this decrease can be described by the slope of the linear least square fit for the pain thresholds over time. The slope was computed for each volunteer under remifentanil and under placebo, respectively, and 95% confidence interval (CIs) are presented below. In addition, the difference between these trends was analyzed by paired t-tests.

Results

Twenty volunteers were studied. Seventeen subjects completed the study. Three volunteers were excluded because of early withdrawal from the study: two for personal reasons before the second session and the third for excessive sedation (sedation score of two). Both study groups were comparable with respect to their demographic data (Table 1). The baseline values of both groups showed no significant differences in thermal or electrical current thresholds.

Table 1: Demographic Data by Sequence Groups

The devolution of the PPTs and PTTs during 180 min of constant infusion of remifentanil or placebo are presented in Figure 1–4. Table 2 gives the mean, se, and 95% CI for the trend of the PPTs and PTTs, respectively (threshold difference between time points three and five), with placebo and remifentanil. All CIs contain the value zero. Therefore, for zero of the pain thresholds, a significant decrease (or increase) could be detected neither with placebo nor with remifentanil. In addition, Table 2 gives the mean, se, and 95% CI for the remifentanil/placebo difference of the trends. No significant difference of the pain threshold devolution at time points three to five between placebo and remifentanil could be detected. There were no significant cardiovascular or respiratory side effects during the entire study.

Figure 1: Heat pain perception threshold (PPT) during constant infusion of 0.08 μg · kg⁻¹ · min⁻¹ of remifentanil (–•–) versus saline (–•–) and heat pain tolerance threshold (PTT, –□– versus –▪–) measured over 180 min. Mean ± sem (°C), n = 17.

Figure 2: Cold pain perception threshold (PPT) during constant infusion of 0.08 μg · kg⁻¹ · min⁻¹ of remifentanil (–•–) and saline (–•–) and cold pain tolerance threshold (PTT; –□– versus –▪–) measured over 180 min. Mean ± sem (°C), n = 17.

Figure 3: Electrical pain perception threshold (PPT) and electrical pain tolerance threshold at 5 Hz (PTT; –□– versus –▪–) during constant infusion of 0.08 μg · kg⁻¹ · min⁻¹ of remifentanil (–•–) and saline (–•–) over 180 min. Mean ± sem (mA), n = 17.

Figure 4: Electrical pain perception threshold (PPT) and electrical pain tolerance threshold at 250 Hz (PTT; –□– versus –▪–) during constant infusion of 0.08 μg · kg⁻¹ · min⁻¹ of remifentanil (–•–) and saline (–•–) over 180 min. Mean ± sem (mA), n = 17.

Table 2: Mean trends (±se) of the pain perception thresholds (EPPT, CoPPT, HPPT) and pain tolerance thresholds (EPTT, CoPTT, HPTT) to current at 5 and 250 Hz and to cold and heat between 55 and 180 minutes after start of a constant infusion of remifentanil or placebo, and mean trend differences remifentanil/placebo. For none of the pain thresholds a significant decrease could be detected neither under remifentanil nor under placebo. No significant difference for the pain threshold devolutions between remifentanil and placebo could be detected.

Discussion

The development of opioid tolerance may be assessed by either increase of the opioid dose to maintain a constant analgesic effect or by decrease of the analgesic effect while the opioid concentration remains constant. In this study, we chose the latter approach, defining tolerance as a decrease of pain thresholds at steady-state of a constant remifentanil infusion.

Opioid tolerance develops over time. A study addressing tolerance may underlay time-dependent effects and has to take into account that experimental pain measurements may be influenced by increasing or decreasing effects such as sensitization or habituation to repeated pain stimuli. Therefore, in our study, the threshold assessments during placebo infusion were included into the analysis. During placebo, no change of thresholds (neither increase nor decrease) was observed within the time course of analysis. In addition, the difference between placebo and remifentanil did not show any significant difference of the pain threshold devolution.

Therefore, the main finding of this study is that there is no change in analgesic effect of remifentanil between 55 and 180 minutes after the start of a constant infusion of remifentanil. In this first placebo-controlled study, we could not demonstrate the development of acute opioid tolerance to remifentanil in a human experimental pain model. There are only a few studies available addressing rapid development of tolerance in humans.

Our results are in agreement with the findings of Schraag et al. ⁽⁴⁾ who found no evidence of tolerance to remifentanil or alfentanil in postoperative patient-maintained, target-controlled infusions of the opioids during 6 h and 24 h, respectively, after surgery.

In contrast, Vinik and Kissin ⁽³⁾ studied the rapid development of tolerance to analgesia during remifentanil infusion in 13 volunteers responding to a series of cold pressor tests. They observed a 75% decrease of the peak analgesic effect after 3 hours of a remifentanil infusion, which was approximately in the same dose range (0.1 μg · kg⁻¹ · min⁻¹) as our study. Two volunteers were excluded who had a positive cold pressor response to a saline infusion 30 minutes before the remifentanil infusion, one volunteer was excluded for lack of a reliable analgesic response, and one volunteer was excluded for repeated reaching of the analgesic cutoff limits. Thus, approximately one-third of the participants of the study were excluded for inappropriate responses to the cold pressor test, implying that the results of this experimental setting may not be reliable.

The cold pressor test is a human pain-induction model, which may be influenced by several confounding factors particularly in a setting of repeated pain stimulation over time. During the cold pressor test complex interactions between neuroendocrine and psychological processes can affect pain appraisal ⁽¹⁷⁾. Staats et al. ⁽¹⁸⁾ showed that suggestion significantly altered participants’ pain perception in the cold pressor test. Moreover, Bayer et al. ⁽¹⁹⁾ demonstrated that previous pain experience in the cold pressor test can influence the reactivity to suggestion for pain. This may explain the different results of Vinik and Kissin ⁽³⁾. In contrast, our study was placebo controlled and could not show any acute opioid tolerance to remifentanil infusion.

Human experimental pain models induce different types of pain. It is their advantage to provide standardization of pain stimuli, but as a disadvantage, they are different from clinical pain. We used quantitative sensory testing, which is well established in the assessment of nociception in humans ^(11,20). In this study, we used two different stimuli. Both thermal and current threshold testing provide standardized and quantitative results. Both methods are sensitive to morphine-mediated analgesia ^(7,21). The electrical current threshold methodology used in this study has also been used to study the effects of epidural and IV fentanyl ⁽²²⁾. Recently, we have shown the dose-dependent effects of remifentanil on thermal pain thresholds ⁽¹⁰⁾.

However, in animals, the development of early opioid-tolerance is well established ^(1,23). Moreover, it develops more rapidly with larger doses ⁽²⁴⁾. The setting of our study is limited to the dose range used because self-administered pain threshold tests are reaction time dependent and require cooperative subjects. We have observed dose-dependent sedative and possible respiratory depressant effects of remifentanil in a previous study ⁽¹⁰⁾ and in one subject of the current study, which may therefore preclude testing larger doses of remifentanil than used in our study using the same methodology.

In a recent human study, Guignard et al. ⁽²⁾ suggested that an increase in morphine consumption after remifentanil-based anesthesia was because of acute development of opioid tolerance. The patients in this study were anesthetized with doses of remifentanil that were four times as much as given to our volunteers. This finding suggests that larger doses of opioids may result in acute opioid tolerance. However, Guignard et al. failed to demonstrate a time-dependent increase in remifentanil requirements during anesthesia, as it would be expected for the development of acute tolerance. They additionally failed to provide a good explanation as to why higher postoperative pain ratings may not be caused by hyperalgesia after stopping large-dose remifentanil infusions than to the development of opioid tolerance. However, our results are restricted to small-dose remifentanil. In conclusion, during continuous small-dose remifentanil infusion, the rapid development of opioid tolerance in volunteers assessed by thermal and electrical current quantitative sensory testing could not be observed.

The authors gratefully acknowledge W. Brannath, Institute of Medical Statistics of the University of Vienna, Austria, for statistical analysis and the Ludwig Boltzmann Institute for Experimental Anesthesiology and Research in Intensive Care Medicine, Vienna, Austria, for technical support.