Comparative Analgesic and Mental Effects of Increasing Plasma Concentrations of Dexmedetomidine and Alfentanil in Humans

The study was approved by the Institutional Review Board of Stanford University. Twelve healthy subjects (eight men, four women), aged 24 ± 2 yr (mean ± SD) and weighing 74 ± 12 kg (men) and 63 ± 8 kg (women), gave written informed consent. All subjects had a normal physical examination, electrocardiogram, and routine laboratory profile, and a negative drug screen and pregnancy test (women) before enrollment. Subjects did not take any medication except oral contraceptives 4 weeks before and during study participation.

Using a double-blind design subjects were randomly allocated to receive intravenous infusions of dexmedetomidine, alfentanil, and saline placebo. All subjects received all treatments but during different study sessions at least 5 days apart (Latin square randomization). Before participation subjects were familiarized with all tests employed during the study. Timing and the number of tests performed during the training corresponded to that employed during a study session.

Before each study day subjects fasted overnight. On arrival at the study center a catheter was inserted in a vein of the left arm for blood drawings. A baseline blood sample was obtained. Another catheter was inserted in a foot vein for drug administration. Monitoring of vital signs was started (electrocardiogram, noninvasive arterial blood pressure, hemoglobin oxygen saturation, and respiratory rate). An intravenous dose of 0.2 mg glycopyrrolate was administered as a precaution to prevent profound bradycardia during drug infusion. Baseline vital signs were recorded 2 min later, followed by baseline pain and mental testing in fixed time order: Experimental heat pain (∼5 min), experimental electrical pain (∼12 min), trail-making test (∼2 min), reaction time (∼2 min), and sedation scores (<1 min). These tests are described in more detail below.

Upon completion of baseline testing a computer controlled drug infusion was started aiming at four geometrically increasing (factor 2) target plasma concentrations. Each target concentration was maintained for 45–60 min. While maintaining a target concentration blood for assaying drug plasma concentration was sampled at 15 and 30 min and at the end of each infusion step. Vital signs were recorded at 15 min and at the end of each infusion step. Experimental pain testing followed by mental testing was started 15 min after initiating each infusion step. No testing was performed during the first 15 min of an infusion step to allow drug equilibration between plasma and effect site. It was therefore reasonable to assume that during the actual testing the drug concentration at the effect site was fairly constant and closely related to the measured plasma concentration.

After termination of the drug infusion two subsequent test cycles were performed that were identical in timing and content to those performed during drug infusion. A 15-min interval without testing separated the end of the infusion and the first postinfusion test cycle and the first and second postinfusion test cycle, respectively.

The α²-adrenergic receptor agonist dexmedetomidine was supplied by Abbott Laboratories (Abbott Park, IL) and the μ-opioid receptor agonist alfentanil was obtained from Janssen Pharmaceutica (Titusville, NJ).

A computer controlled infusion pump (Harvard Pump 22; Harvard Apparatus Inc., South Natick, MA) was used to rapidly achieve and maintain steady state plasma concentrations. Target concentrations increased geometrically and were 0.1, 0.2, 0.4, and 0.8 ng/ml for dexmedetomidine and 20, 40, 80, and 160 ng/ml for alfentanil, respectively. Dexmedetomidine target plasma concentrations were selected to study analgesic effects to experimental pain at concentrations producing mild to severe sedation but not unconsciousness.7Alfentanil target plasma concentrations were selected to test the sensitivity of the employed experimental pain model at concentrations providing postoperative pain control and similar levels of sedation as dexmedetomidine.9STANPUMP§was the software driving the pump infusion. The weight-adjusted pharmacokinetic parameters used with STANPUMP are listed in table 1. The person operating the computer-controlled infusion was not blinded to the treatment but did not interact with the study volunteer or the investigator conducting pain and mental tests.

Five milliliters of venous blood were drawn into heparinized glass tubes, centrifuged, frozen, and stored at −20°C. Dexmedetomidine plasma concentrations were determined by negative ion gas chromatography—mass spectroscopy at Oneida Research Services, Inc. (Whitesboro, New York). Alfentanil plasma concentrations were determined using an electro-spray liquid chromatography tandem mass spectrometry method at Alta Analytical Laboratories, Inc. (El Dorado Hills, CA). The lower limits of quantification were 10 pg/ml for dexmedetomidine and 0.92 ng/ml for alfentanil. The coefficient of variation across a dexmedetomidine concentration range of 50–1500 pg/ml was 4.0–5.4% and across an alfentanil concentration range of 3–345 ng/ml was 3.2–6.0%.

Nociceptive heat and electrical stimuli were used to test for analgesic effects before, during, and after administration of dexmedetomidine, alfentanil, and saline placebo. The lowest temperature evoking pain (pain threshold) and the highest temperature tolerated (pain tolerance) were determined using a small metal plate in contact with skin. The lowest current evoking pain and the highest current tolerated were determined by constant current administration via  a skin-surface electrode. Standardized sentences describing the procedure and defining the measured end points were read to subjects before starting an experimental pain test.

As reported previously, a thermal sensory analyzer (TSA 2001; Medoc Advanced Medical Systems, Minneapolis, MN) was used to administer the nociceptive heat stimuli.11In brief, a hand-held 16 × 16 mm thermode was brought into full contact with skin at the right forearm. After equilibration between skin and thermode temperature at 35°C, the temperature of the thermode was increased by 1°C/s (cutoff at 53°C). Subjects pushed the button of a hand-held device as soon as they felt pain, thereby triggering the recording of the temperature causing pain as well as the immediate cooling of the probe. This procedure was performed three times at sites at least 5 cm apart, and the average of the minimal temperature evoking pain was recorded. Next, using the same algorithm subjects pushed the button of the hand-held device as soon as they were unable to tolerate the evoked pain. The average of the maximum tolerated temperature was recorded. The interstimulus interval was 30 s. If a subject was able to tolerate the maximum output temperature of the device (53°C), this was recorded as the maximum tolerated temperature (16 of 756 recordings).

A constant current device (Neurometer; Neurotron Inc., Baltimore, MD) with a maximum output of 20 mA and delivering 5 Hz sine wave pulses of 3-s duration was used to administer nociceptive electrical stimuli. An aluminum/gold electrode was attached to the surface of the skin at the right lateral upper arm as previously described.11The interstimulus interval was 15 s.

The test algorithm (slightly modified) and the derivation of the pain threshold and the pain tolerance have been illustrated and described in detail elsewhere.12Briefly, a first series of stimuli of increasing intensity was administered to estimate a subject’s pain threshold and pain tolerance. If a subject was able to tolerate the maximum output current of the device (20 mA) this was recorded as the maximum tolerated current (six of 252 series). A second series of 10 stimuli of different intensities evenly spaced between the pain threshold and the pain tolerance was administered in random sequence. Subjects rated the magnitude of pain evoked by each stimulus on a 100-mm visual analog scale anchored by the words “no pain” and “most intense pain tolerable.” A scatter plot depicting the 10 stimulus intensities (mA) versus  pain (visual analog scale) resulted. The data were fitted with a linear model (y = a × (x − b), where x is the current in mA and y is the visual analog scale pain score). The analgesic efficacy variable was the pain threshold determined as the x-intercept and the pain tolerance calculated with aid of the liner model by setting the visual analog scale score to 100.

A modified trail-making test (originally published in German as the “Zahlen-Verbindungs-Test” or “ZVT”) was used to assess cognitive speed. The trail-making test is considered a sensitive measure of cognitive performance and correlates significantly with some tests assessing intelligence.13,14This paper-and-pencil test consists of four matrices featuring 90 numbers organized in nine rows and 10 columns on a 23 × 21 cm sheet of paper. Subsequent numbers are located in a neighboring row or column. Starting at number 1, a subject has to connect subsequent numbers as quickly as possible. The time to complete the task is recorded. The particular matrix a subject had to complete during a test cycle was chosen randomly.

The reaction time was measured to assess a subject’s alertness and particularly the ability to enhance and sustain reactivity while expecting a high priority signal.15This test is considered a sensitive measure of cognitive performance. Subjects were asked to push the key of a hand-held device as quickly as possible when a dot presented on a computer screen changed to a cross. An acoustic signal alerted subjects that the change was imminent. The time interval between the acoustic signal and the change varied randomly. The reaction time was measured 20 times per test cycle and the median was recorded.

Subjects indicated how sedated they felt by setting a mark on a 100-mm visual analog scale relative to a left and right sided verbal anchor using the wording “not at all” and “as much as possible” (0 = not sedated at all, 100 = sedated as much as possible).

Results summarized in tables are expressed as mean and SD if data passed the normality test (Kolmogorov-Smirnov) or alternately as median and interquartile range. However, the SEM, rather than the SD, is represented by error bars to facilitate the reading of graphs.

To determine whether plasma concentrations changed significantly during an infusion step the concentrations measured at 15 and 30 min, and at the end of an infusion step were compared among each other using parametric or nonparametric one-way repeated measures analysis of variance and Student-Newman-Keuls post hoc  test. As there were four infusion steps, a P  value < 0.0125 was considered statistically significant (Bonferroni correction).

To compare analgesic effects, sedative effects, and changes in vital signs among treatments with dexmedetomidine, alfentanil, and saline placebo individual areas under the curves depicting the time versus  the effect measure were calculated using linear interpolation. Parametric or nonparametric one-way repeated measures analysis of variance and Student-Newman-Keuls test were used to detect significant differences among treatments.

If administration of dexmedetomidine or alfentanil resulted in an effect that was significantly different from that observed for saline placebo, regression analysis was used in pooled data to explore whether a significant plasma concentration versus  effect relationship existed. The plasma concentration measured at 15 min after starting each infusion step was used. Plasma concentrations measured during the washout phase of a drug were not considered because equilibration between plasma and effect site could not be assumed precluding the direct correlation of an effect with an actual plasma concentration. Employed pharmacodynamic models were nested in a power model and a maximum likelihood approach was used for parameter estimation. Model equations are not listed in the manuscript because the scope of this study was to determine whether a significant plasma concentration versus  effect relationship existed for a particular effect measure but not to establish pharmacodynamic models allowing accurate predictions and extrapolations. A P  value < 0.05 was considered statistically significant.

All 12 subjects completed the investigation according to the protocol. There were no unexpected adverse events. One subject received 0.2 mg glycopyrrolate intravenously for bradycardia (<40 beats/min) during the infusion of dexmedetomidine. Another subject received 10 mg metoclopramide intravenously for severe nausea during the infusion of alfentanil. No other medications (except oral contraceptives and study drug) were administered during study participation.

Steady state target plasma concentrations were stable during the infusion steps except for a slight but significant increase of the alfentanil plasma concentration during the first step. Target plasma concentrations, measured plasma concentrations, and cumulative drug doses are listed in table 2. The median coefficient of variation of the steady state plasma concentration was 13% for dexmedetomidine (range, 1 to 55%) and 12% for alfentanil (range, 2 to 77%). The median measured plasma concentration of dexmedetomidine was 13% (range, −100 to 87%) higher than the target concentration. The median measured plasma concentration of alfentanil was 20% lower (range, −84 to 27%) than the target concentration.

Two subjects receiving dexmedetomidine were too sedated to complete experimental pain testing during the fourth infusion and the first postinfusion test cycle.

Figure 1(upper graphs) depicts the change of the heat pain threshold and the heat pain tolerance from baseline during and after intravenous infusion of dexmedetomidine and alfentanil. Table 3lists the corresponding values as absolute numbers (°C). Dexmedetomidine was similar to placebo administration and had no analgesic effects irrespective of measured plasma concentration. However, administration of alfentanil resulted in significant analgesia, as indicated by an increased heat pain threshold and heat pain tolerance compared with placebo and dexmedetomidine administration. The increase in pain tolerance, but not pain threshold, was dependent on the plasma concentration. This relationship was linear within the range of plasma concentrations explored in this study. As demonstrated previously, the pain tolerance is more precise than the pain threshold for describing opioid-induced analgesic effects.11 

Figure 1(middle graphs) depicts the percentage change of the electrical pain threshold and electrical pain tolerance from baseline during and after intravenous infusion of dexmedetomidine and alfentanil. Table 3lists the corresponding values as absolute numbers (mA). Dexmedetomidine was similar to placebo administration and had no analgesic effects irrespective of measured plasma concentration. However, administration of alfentanil resulted in a plasma concentration dependent increase in pain threshold and pain tolerance that was significantly different form saline placebo (pain tolerance only) and dexmedetomidine administration (pain tolerance and pain threshold). The relationship between the plasma concentration and the effect measures was linear within the range of concentrations explored in this study.

Compared with placebo administration dexmedetomidine and alfentanil decreased the speed of cognitive performance significantly and in a plasma concentration dependent fashion (fig. 2). The relationship between the plasma concentration and the effect measure was linear within the range of concentrations explored in this study. Table 4lists the corresponding data as absolute numbers. No significant difference was detected between dexmedetomidine and alfentanil administration. At the highest plasma concentration the time to complete the trail-making test was increased by a factor of 2.0 for dexmedetomidine and of 1.4 for alfentanil. Based on data demonstrating a significant correlation between the performance in the trail-making test and the intelligence quotient, the average cognitive speed of our study population corresponded to an intelligence quotient of 122 before drug administration but corresponded to intelligence quotients of 75 and 97 while exposed to the highest dexmedetomidine and alfentanil plasma concentrations.13 

Compared with placebo administration dexmedetomidine and alfentanil resulted in a significant and plasma concentration dependent increase of the reaction time (fig. 2). Within the range of plasma concentrations explored in this study, the relationship between the concentration and the effect measure was exponential (exponent > 1) for dexmedetomidine and linear for alfentanil. Table 4lists the corresponding data as absolute numbers. The reaction time increased significantly more during dexmedetomidine than during alfentanil infusion, i.e. , by a factor of 1.8 and 1.4 at the highest plasma concentrations. Based on normative data the average reaction time of our study population corresponded to the 80th percentile before drug infusion but dropped down to the 2nd and 10th percentiles while subjects were exposed to the highest plasma concentrations of dexmedetomidine and alfentanil.15 

Compared with placebo administration dexmedetomidine and alfentanil increased the sedation score significantly and in a plasma concentration dependent fashion (fig. 2). Within the range of plasma concentrations explored in this study, the relationship between the concentration and the effect measure was exponential (exponent < 1) for dexmedetomidine and linear for alfentanil. Table 4lists the corresponding data as absolute numbers. No significant difference was detected between dexmedetomidine and alfentanil administration. Sedation scores tended to be higher during dexmedetomidine infusion and subjects indicated to be almost maximally sedated at the highest plasma concentration. Sedation scores before administering dexmedetomidine, alfentanil, and placebo ranged between 22 and 28, indicating that a portion of the overall sedation score was not related to drug administration.

Infusion of dexmedetomidine resulted in a significant decrease of systolic and diastolic blood pressures and heart rate as compared with alfentanil and saline placebo (fig. 3). The relationship between the plasma concentration and the three effect measures was exponential (exponent <1) within the range of concentrations explored in this study. No difference was detected between alfentanil and saline placebo administration.

Infusion of alfentanil resulted in a significant decrease of respiratory rate and hemoglobin oxygen saturation as compared with dexmedetomidine and saline placebo (fig. 3). The relationship between the plasma concentration and the two effect measures was linear within the range of concentrations explored in this study. No difference was detected between dexmedetomidine and saline placebo administration.

The aim of this study in human volunteers was to determine whether systemic administration of the α²-adrenergic receptor agonist dexmedetomidine attenuates experimentally induced heat and electrical pain at plasma concentrations, resulting in mild to severe sedation. We report that intravenous administration of dexmedetomidine lacked analgesic efficacy, even at plasma concentrations causing severe sedation and impairment of cognitive speed.

Systemic administration of the α²-adrenergic receptor agonists dexmedetomidine and clonidine has been reported to produce sedative and opioid-sparing effects in the perioperative setting, providing indirect evidence for some analgesic efficacy.16–20However, studies conducted in the perioperative setting are subjected to confounding factors and it is difficult to distinguish between analgesic and sedative effects as a cause for observed opioid-sparing effects.

Experimental pain studies in human volunteers using the cold pressor test documented a 20–30% decrease of the visual analog pain score in subjects receiving dexmedetomidine or clonidine at doses causing moderate to severe sedation.6,7,21In contrast, a study performed with experimental heat pain and electrical tooth pulp stimulation failed to detect any analgesic effect while reporting a significant attenuation of the unpleasantness, but not of the intensity, of pain in a model of ischemic pain after administration of the α^2-adrenergic receptor agonist medetomidine at a dose producing severe sedation.8Finally, volunteers exposed to clonidine at a dose producing moderate to severe sedation did not experience any antihyperalgesic or antiallodynic effect in an experimental model of secondary hyperalgesia.22Our results add to the evidence suggesting that dexmedetomidine lacks broad analgesic efficacy after systemic administration of doses producing mild to severe sedation.

Experimental pain may not predict whether a systemic dose of dexmedetomidine can alleviate clinical pain. For example, systemic administration of medetomidine attenuates the unpleasantness, but not the intensity, of ischemic pain.8Reducing the unpleasantness of pain may be relevant in the treatment of clinical pain. Similarly, dexmedetomidine interacts synergistically with endogenous mechanisms underlying stress-induced analgesia and therefore may have a role as an analgesic compound for clinical pain associated with high levels of stress.23 

However, there is robust evidence that the α²agonist clonidine administered via  the intrathecal or epidural route produces significant analgesia in postoperative pain and in cancer pain states, for the latter particularly if neuropathic in origin.24–26Interestingly, experimental pain models that failed to detect an analgesic effect of a α²agonist after its systemic administration, readily detect such an effect after its intrathecal or epidural administration.27Parallel documentation of analgesic efficacy in models of clinical and experimental pain for the intrathecal and epidural route, and lack thereof for the intravenous route, casts further doubt on whether systemic administration of a α²-agonist will provide a robust analgesic effect for a variety of clinical pain states. This does not preclude the possibility that systemic administration of a α²agonist could be useful for some types of clinical pain or may be a valuable therapeutic adjuvant acting synergistically to enhance the analgesic action of another pain therapeutic.

Animal data show a dose-dependent response for both antinociception and sedation with systemic administration of a α²agonist.28,29Human data show a clear dose-response relationship for sedation, but not for analgesia, with systemic administration of a α²agonist.8,30,31Findings between animal and human studies may be divergent because animal studies used doses that were several orders of magnitude larger than those used in human trials.18,28–30Human studies may not allow using an effective analgesic dose of a α²agonist via  the systemic route because such a dose may render subjects heavily sedated or even unconscious. Administration of a α²agonists via  the intrathecal or epidural route likely spares supraspinal central nervous system sites from extensive drug exposure, thereby providing robust analgesic effects in most subjects without producing severe sedation.28,30 

Despite profound sedation and substantial impairment of cognitive speed at high plasma concentrations of dexmedetomidine and alfentanil, dexmedetomidine exerted no analgesic action, while alfentanil resulted in a robust analgesic effect. Contrariwise, significant analgesic effects of alfentanil were documented at low plasma concentrations that did not produce profound sedation or mental impairment. This study illustrates that sedative and analgesic drug effects could clearly be distinguished, i.e. , did not confound each other.

Dexmedetomidine decreased heart rate and blood pressure in dose-dependent fashion. These cardiovascular effects of dexmedetomidine are well documented for the explored plasma concentrations.7However, at higher plasma concentrations dexmedetomidine increases blood pressure while heart rate is further decreased.7Dexmedetomidine did not affect respiratory rate or hemoglobin oxygen saturation. This is consistent with previous data suggesting that ventilation is only mildly affected at plasma concentrations up to 10 times higher than the peak concentrations reported here.7However, if dexmedetomidine is administered intravenously as a bolus (1–2 μg/kg in 2 min) the ventilatory pattern can become irregular and short episodes of apnea have been described.5 

In summary, this study provides further evidence in human volunteers that systemically administered dexmedetomidine lacks broad analgesic activity in models of acute pain at plasma concentrations producing mild to severe sedation but not unconsciousness.