Lumbar Epidural Morphine in Humans and Supraspinal Analgesia to Experimental Heat Pain

Nine healthy volunteers gave written informed consent for the study, which was approved by the Institutional Review Board of Stanford University. All subjects had a negative physical examination, an unremarkable medical history, and did not take any medications.

This study used a randomized, double-blinded, placebo-controlled, crossover design. Either morphine or normal saline was injected into the epidural space during two sessions at least 6 days apart. Study sessions started at 7 AM in an isolated and quiet room at an ambient temperature that was comfortable to the subjects. An intravenous catheter was placed in one arm for blood drawings. Subjects remained semirecumbent throughout the study.

An 18-gauge Touhy needle was inserted midline via  the L2/L3 or L3/L4 interspace. Subjects were in the right lateral position. The epidural space was identified by loss of resistance to air. Either 5 mg preservative-free morphine (Abbott Laboratories, North Chicago, IL) diluted in 10 ml normal saline or 10 ml plain normal saline was injected over a period of 60 s via  syringe pump (Harvard Pump 22; Harvard Apparatus Inc., South Natick, MA). The investigator who prepared and administered the drug did not take any further part in the study. Immediately after drug administration, an 18-gauge epidural catheter was inserted through the needle up to 5 cm into the epidural space. Five hours after epidural drug administration, 10 ml radio-opaque dye (Omnipaque; Nycomed Inc., Princeton, NJ) was injected through the epidural catheter. Fluoroscopy confirmed the correct placement of the epidural catheter.

Nociceptive heat and electrical stimuli were used to test for analgesic effects after administration of epidural morphine or 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 using a skin-surface electrode. Recordings were made at baseline and 2, 5, 10, and 24 h after drug administration. During each analgesic measurement period, three different dermatomes were evaluated in random sequence. The lumbar and thoracic dermatomes tested for nociceptive heat and electrical stimuli were identical. However, the most cranial dermatomes tested for nociceptive heat and electrical stimuli were different. The cheek was used to test as far rostral as possible for nociceptive heat stimuli. This or a nearby facial location could not be used for nociceptive electrical stimuli because such stimulation not only evoked pain but also muscular twitching proportional in strength to underlying stimulus intensity (not observed for lumbar and thoracic location). The perceived magnitude of a muscular twitch provides a clue about the intensity of administered nociceptive stimulus and in this way may bias and invalidate how subjects rate the magnitude of evoked pain. Therefore, the pinna of the ear was used as the most rostral dermatome feasible for electrical stimulation. To familiarize subjects with the test procedure, they all underwent two sessions of experimental pain testing a week before the study.

A thermal sensory analyzer (TSA 2001; Medoc Advanced Medical Systems, Minneapolis, MN) was used to administer the nociceptive heat stimuli. An investigator brought a hand-held 16 × 16-mm thermode in full contact with the subject’s skin. After equilibration between skin and the thermode at 35°C, the temperature of the thermode increased 1°C/s. The thermode could be heated to a maximum temperature of 53°C. First, subjects pushed the button of a hand-held device as soon as they felt pain, thereby triggering the recording of the temperature that caused pain as well as the immediate cooling of the probe. This procedure was performed three times per dermatome at sites at least 2.5 cm apart, and the average of the minimal temperature evoking pain was recorded. 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 (24 of 810 recordings). The dermatomes evaluated for nociceptive heat stimuli were L4 (lateral to the tibia), T10 (lateral to the umbilicus), and V2 (cheek). At the beginning of each test cycle, standardized sentences were read to the subjects emphasizing the algorithm of the procedure.

A constant-current device (Neurometer; Neurotron Inc., Baltimore, MD) with a maximum output of 20 mA and delivering 5-Hz sine wave pulses was used to administer nociceptive electrical stimuli. A ring electrode (outer aluminum electrode 16 mm wide and concentric to a central gold electrode 5 mm in diameter; distance between adjacent margins was 14 mm) was attached to the surface of the skin. The electrical pain threshold was determined by delivering at first a non-noxious stimulus randomly ranging between 80 and 160 μA. The second stimulus was 50% higher than the first one. The magnitude of subsequent stimuli was determined by a subject’s response to the two preceding stimuli. If a subject’s response to two preceding stimuli was “no pain—no pain” or “pain—pain” the next stimulus was equal to the magnitude of the last delivered stimulus plus or minus 130% of the difference between the last and the second last stimulus, respectively. If a subject’s response to two preceding stimuli was “no pain—pain” or “pain—no pain,” the next stimulus was equal to the magnitude of the second last stimulus plus 75% or 25% of the difference between the last and the second last stimulus, respectively. The purpose of outlined algorithm was to increase or decrease the magnitude of delivered stimuli quickly as long as a subject gave a uniform response, i.e. , “no pain” or “pain.” However, whenever a subject changed the response from “no pain” to “pain” or from “pain” to “no pain,” the outlined algorithm allowed exploring the magnitude of stimuli evoking a change in response at a higher resolution. Once the value of the highest current evoking no pain and the lowest current evoking pain deviated by no more than 10% from each other, their average was recorded as the pain threshold. The electrical pain tolerance was determined by delivering at first a stimulus randomly ranging between 160 and 240 μA. The magnitude of each subsequent stimulus was increased by 15%. The stimulus series stopped as soon as a subject indicated that the next higher stimulus would induce intolerable pain. None of the subjects reached the maximum output of the constant current device. This procedure was performed twice per dermatome, and the average of the maximum-tolerated current was recorded. The stimulus duration was 3 s, and the interstimulus interval was 15 s. The dermatomes evaluated were L4 (lateral to the tibia), T10 (paraspinal), and C2 (pinna of the ear with a modified ring electrode). At the beginning of each test cycle, standardized sentences were read to the subjects, emphasizing the algorithm of the procedure.

Five milliliters of venous blood was collected into a heparinized tube to determine the plasma morphine concentration before and 2, 3, 5, 6, 10, 11, 24, and 25 h after epidural drug injection. Samples were immediately put on ice, centrifuged within 2 h, and stored at −20°C.

The assay for morphine was conducted at the Bioanalytical Laboratory of the Pain Research Program at Fred Hutchinson Cancer Research Center, Seattle, Washington. Morphine plasma concentrations were determined with an HP 5890II gas chromatograph and HP 5989A mass spectrometer (Agilent Technologies, Palo Alto, CA). Nalorphine as internal standard (20 ng), boric acid/sodium borate buffer (1 ml; pH = 8.9), and a chloroform isopropyl alcohol mixture (4 ml; ratio 95:5) were added to each sample (0.5 ml). Samples were shaken (150 rpm) for 15 min and centrifuged for 10 min (3,000 rpm). The aqueous layer was removed by suction. The organic layer was evaporated to dryness under a stream of nitrogen (65°C). After cooling to room temperature, 50 μl pentafluoropropionic anhydride was added and the tubes were immediately capped and heated to 65°C for 45 min. Pentafluoropropionic anhydride was evaporated under a stream of nitrogen at room temperature. The residue was reconstituted in ethyl acetate (100 μl) and injected on the gas chromatograph/mass spectrometer (1–2 μl). Standard curves were prepared daily for concentrations between 2 and 400 ng/ml. Spiked control samples had concentrations of 30 ng/ml and 150 ng/ml, respectively. The within-day coefficients of variation were 9.2 and 10.9%, and the between-day coefficients of variation were 10.6 and 12.6%, respectively. The limit of quantification was 1 ng/ml.

Blood pressure, heart rate, respiratory rate, and hemoglobin oxygen saturation were assessed noninvasively at baseline, every 15 min during the first 2 h after drug administration, and then hourly up to 24 h. Adverse events were recorded.

To determine drug effect on the minimum thermode temperature evoking pain (heat pain threshold) and the maximum-tolerated thermode temperature (heat pain tolerance), individual readings before epidural drug administration were subtracted from readings obtained 2, 5, 10, and 24 h after drug administration. The difference from baseline was used because temperature is measured on an interval scale that precludes expressing a change in temperature as a percentage. The area under the curve (AUC) describing differences of the thermode temperature versus  time was calculated for each subject. Six individual AUCs describing the effect of epidural morphine or saline on the pain threshold and the pain tolerance at L4, T10, and V2 were derived, respectively. The AUC was calculated by the trapezoidal rule using linear interpolation.

To determine drug effect on the minimum current evoking pain (electrical pain threshold) and the maximum-tolerated current (electrical pain tolerance), individual readings obtained at 2, 5, 10 and 24 h after epidural drug administration were expressed as the percentage change from the reading recorded before epidural drug injection. The AUC describing the percentage change of administered current versus  time was calculated as outlined previously.

The statistical analysis aimed to address two questions. First, at which dermatomes does morphine produce significant analgesia when compared with saline placebo? Second, at what times after epidural injection was this effect significant compared with saline placebo? To address the first question, individual AUCs depicting the pain threshold or the pain tolerance versus  time after epidural morphine and saline placebo injection were compared (paired t  test or Wilcoxon signed rank test). Because such comparison was made at three dermatomes, a P  value of 0.017 (Bonferroni’s correction) was considered statistically significant. If epidural morphine produced significant analgesia at a particular dermatome, morphine and saline placebo treatments were compared at 2, 5, 10, and 24 h after epidural injection. A P  value of 0.0125 (Bonferroni’s correction) was considered statistically significant because four comparisons were made. The paired test procedure (paired t  test or Wilcoxon signed rank test) was preferred to a three-factorial analysis of variance because not all of the pain tolerance and pain threshold data were normally distributed, nor was the sphericity assumption implicit to the analysis of variance supported. All data are presented as mean and SD unless otherwise stated.

All of the five male and four female subjects (age 26 ± 5 yr, weight 55 to 96 kg) completed the study.

The epidural space was identified in all subjects on the first attempt. After placement of the epidural needle, all attempts to aspirate fluid were negative. Correct placement of the catheter inserted through the epidural needle after drug administration was confirmed by fluoroscopy of the lumbar spine. The radio-opaque dye injected showed the typical column-like spread within the epidural space reaching a rostral level of L1 to T10 after the 60-s injection period.

Table 1lists the raw values of the heat pain threshold and the heat pain tolerance obtained at L4, T10, and V2 before and after epidural administration of morphine and saline placebo, respectively. In contrast, figures 1 and 2depict individual changes of the heat pain threshold and the heat pain tolerance from baseline at L4, T10, and V2 after morphine (top graphs) and saline placebo (bottom graphs), respectively. Figure 3summarizes the individual data presented in figures 1 and 2, depicting the average change of the heat pain threshold (top graphs) and the heat pain tolerance (bottom graphs) from baseline.

The heat pain threshold increased at all dermatomes after epidural morphine but did not change or increased just slightly after epidural saline injection. The heat pain threshold peaked between 5 and 10 h and remained elevated during the 24-h observation period. Changes were more pronounced at L4 and T10 than at V2. The AUC describing the difference of the heat pain threshold from baseline versus  time was significantly different between epidural morphine and saline injections at L4 and T10 (P < 0.017). The AUC describing the difference of the heat pain threshold from baseline versus  time at V2 was not significantly different but tended to be larger after epidural injection of morphine (P = 0.12). Comparing epidural morphine with epidural saline injections, the heat pain threshold was significantly increased at L4 and T10 after 2, 5, and 10 h and after 5, 10, and 24 h, respectively.

The heat pain tolerance increased at all dermatomes after epidural morphine but did not change or decreased after epidural saline injection. The heat pain tolerance peaked between 5 and 10 h and remained elevated during the 24-h observation period. Changes were more pronounced at L4 and T10 than at V2. The AUC describing the difference of the heat pain tolerance from baseline versus  time was significantly different between epidural morphine and saline injections at all dermatomes tested (P < 0.017). Comparing epidural morphine with epidural saline injections, the heat pain tolerance was significantly increased at L4 after 2, 5, and 10 h; at T10 after 5, 10, and 24 h; and at V2 after 10 h.

Inspection of figure 3suggests that epidural morphine increased the heat pain threshold more profoundly than the heat pain tolerance. However, the variance associated with the heat pain threshold was larger than that associated with the heat pain tolerance (compare magnitude of error bars in fig. 3and SDs in table 1). To determine as to which of the two effect measures (pain threshold or pain tolerance) was more suitable to detect morphine-induced analgesia, the detected difference between treatments (morphine vs.  placebo) needs to be related to associated variance. In other words, not the signal itself (difference between treatments) but the ratio between the signal and associated noise (variance unrelated to treatments) determines how readily an effect measure detects a difference between treatments. One way of relating the so-defined signal to associated noise offers omega-squared statistics. 16Omega-squared estimates the fraction of the overall variability of the effect measure that can be attributed to the fact that subjects received different treatments (morphine vs.  saline placebo). Omega-squared associated with the heat pain threshold and the heat pain tolerance were 0.35 and 0.47 at L4, 0.44 and 0.56 at T10, and 0.14 and 0.39 at V2, respectively. It has been suggested that an omega-squared value > 0.15 indicates a large treatment effect. 16Despite a larger difference between the means of the heat pain threshold after epidural morphine and saline placebo administration, the heat pain tolerance was consistently associated with a larger omega-squared, i.e. , was more suitable to detect morphine-induced analgesia. This explains why, at the level of V2, the heat pain tolerance was significantly different, but the heat pain threshold was only elevated as a trend when comparing epidural morphine with saline placebo administration.

Table 2lists the raw values of the electrical pain threshold and the electrical pain tolerance obtained at L4, T10, and C2 before and after epidural administration of morphine and saline placebo, respectively. In contrast, figures 4 and 5depict individual percentage changes of the electrical pain threshold and the electrical pain tolerance from baseline at L4, T10, and C2 after morphine (top graphs) and saline placebo (bottom graphs), respectively. Figure 6summarizes the individual data presented in figures 4 and 5, depicting the average percentage change of the electrical pain threshold (top graphs) and the electrical pain tolerance (bottom graphs) from baseline.

The electrical pain threshold increased at L4 and T10 but hardly changed at C2 after epidural morphine administration. The electrical pain threshold did not change after epidural injection of saline placebo. The electrical pain threshold peaked between 5 and 24 h. Comparing epidural injection of morphine with saline placebo, the AUC describing the percentage change of the electrical pain threshold from baseline versus  time tended to be larger at L4 and T10 (P = 0.10 and 0.02, respectively) but not at C2.

The electrical pain tolerance increased at L4 and T10 but hardly changed at C2 after epidural morphine administration. The electrical pain tolerance did not change or decreased after epidural injection of saline placebo. The electrical pain tolerance peaked between 2 and 10 h. The AUC describing the percentage change of the electrical pain tolerance from baseline versus  time was significantly different between epidural morphine and saline injections at L4 and T10 but not at C2 (P < 0.017). Comparing epidural morphine with epidural saline injections, the electrical pain tolerance was significantly increased at L4 after 2, 5, and 10 h and at T10 after 5 and 10 h.

Omega-squared associated with the electrical pain threshold and the electrical pain tolerance was determined as outlined for the heat pain data. However, although the same algorithm was used to determine the heat pain threshold and the heat pain tolerance, different algorithms were used to determine the electrical pain threshold and the electrical pain tolerance. This limits interpreting as to which of the electrical pain measures may generally be more suitable to detect morphine-induced analgesia. The electrical pain threshold and the electrical pain tolerance were associated with omega-squared of 0.23 and 0.30 at T10 and 0.00 and 0.04 at C2, respectively. The omega-squared at L4 was 0.40 for the electrical pain tolerance but could not reliably be determined for the electrical pain threshold because of the skewed distribution of the data.

Figure 7shows the average morphine plasma concentration versus  time after epidural injection. Concentrations were highest after 2 h (first measurement), declining exponentially to less than the detection limit (1 ng/ml) in eight subjects after 10 h.

After epidural administration of morphine, seven subjects felt nauseated, three vomited, and seven had difficulties voiding during the first 10 h. All subjects reported pruritus up to 24 h. The time course and incidence of side effects are shown in figure 8. After epidural administration of saline, one subjected felt nauseated.

Changes in blood pressure, heart rate, respiratory rate, and blood oxygen saturation over time were not different after epidural morphine or saline injection. Blood pressure and respiratory rate remained within 15%, and heart rate remained within 25% of the baseline value. Changes were most pronounced at night when subjects were asleep.

This study reports long-lasting supraspinal analgesic effects to nociceptive heat stimuli after epidural administration of 5 mg morphine, but not after epidural injection of saline placebo in the same human volunteers. Significant supraspinal analgesic effects were measured up to 10 h and persisted as a trend for 24 h.

Only few experimental studies have explored the segmental distribution of analgesic effects after lumbar epidural administration of morphine. Bromage et al.  9described a time-dependent analgesic spread to pin-prick–evoked pain up to supraspinal levels after epidural administration of 10 mg morphine. Other investigators reported a significant, time-dependent analgesic spread to laser-evoked pricking pain up to thoracic but not cervical levels after epidural injection of 4 mg morphine. 12Finally, two studies reported analgesic effects to pressure-evoked pain at the forehead after lumbar and low thoracic administration of 3–7 mg morphine. 10,11However, these analgesic effects were only measured up to 3 h, possibly too short a duration to exclude an uptake into the systemic circulation with subsequent distribution of morphine to the brain as an underlying mechanism. Morphine enters the systemic circulation to a similar extent after epidural and intramuscular injection. 13–15 

In our study, 5 mg lumbar epidural morphine resulted in prolonged supraspinal analgesia to heat pain. In contrast, previous investigations reported that 10 mg lumbar epidural morphine attenuated pain to pin prick up to the supraspinal level, but 4 mg lumbar epidural morphine reduced laser-induced pricking pain only up to the thoracic level. 9,12In other words, epidural morphine seems to attenuate heat pain more potently or up to a higher dermatomal level than pricking pain. Behavioral and electrophysiologic evidence suggests that morphine blocks C-fiber–mediated pain more potently than Aδ-fiber–mediated pain. 17–20Behavioral and electrophysiologic evidence also suggests that heat pain is predominantly mediated by C fibers if skin is heated at a slow rate (< 2°C/s). 17,21–24On the other hand, pricking pain is thought to be transmitted primarily by Aδ-fibers. 25–27Therefore, one may speculate that C fibers primarily signaling heat pain were more potently blocked by epidural morphine than Aδ fibers signaling pricking pain, and, consequently, heat pain was attenuated up to a higher dermatomal level than pricking pain.

In a clinical context, the presented finding implies that some types of pain may be attenuated up to the supraspinal level after lumbar epidural administration of morphine. The clinical report of successfully attenuating cancer pain in the head and neck region with a lumbar intrathecal infusion of morphine in a selected group of patients is noteworthy. 28All of these patients suffered from intolerable side effects during previous treatment with systemic opioids.

In our study, significant analgesia was detected for heat pain at the level of the trigeminal nerve but not for electrical pain at a cervical level. In contrast to nociceptive heat consistently evoking a burning pain, electrical pain evoked a mixed pain perception. Subjects reported a sharp or stinging as well as a burning pain component. This indicates simultaneous recruitment of Aδ and C fibers in nociceptive signaling. 24–27The fact that nociceptive electrical stimuli simultaneously stimulated various types of nerve fibers yielding a different sensitivity to morphine may explain why electrical pain was less potently blocked than heat pain. 17–20 

Alternatively, the cervical dorsal horn may not be as sensitive to opioid action as the spinal tract of the trigeminal nucleus involved in processing nociceptive information. After spinal administration of opioids, pruritus, which is reversible by naloxone, is particularly common in the face, i.e. , skin areas innervated by the trigeminal nerve. 29This may point to an exquisite sensitivity of the trigeminal nucleus to opioid action. However, the spinal dorsal horn and the trigeminal spinal tract nucleus share a similar neuroanatomic organization, and both are rich in μ-opioid receptors. 30,31Future research may clarify if opioid potency varies for different spinal and medullary dermatomes.

Analgesic effects to nociceptive heat stimuli were observed at all tested dermatomes 2 h after administration of epidural morphine, peak effects were recorded between 5 and 10 h, and analgesia lasted the entire 24-h observation period. Highest morphine plasma concentrations were measured 2 h after administration of epidural morphine and were approximately half of the minimal effective plasma concentration necessary for the treatment of postoperative pain. 32Plasma morphine concentrations decreased to less than the detection limit (1 ng/ml) in eight subjects after 10 h. Therefore, the long-lasting analgesic action must have been caused by intraspinal mechanisms and cannot be explained by an initial uptake of morphine into the systemic circulation with subsequent distribution to the brain. Analgesic effects peaking between 2 and 7 h and lasting between 13 and > 24 h have been reported after epidural administration of morphine and were attributed to its spinal action. 13–15,33 

Traditionally, a pharmacokinetic explanation is given for effects involving the brain stem after epidural administration of morphine. Morphine is absorbed slowly from the epidural space but then quickly spreads rostrally and becomes detectable in cervical cerebrospinal fluid after 60 min. 6,7,33–35Significant morphine concentrations in cerebrospinal fluid at the brain stem are considered to cause side effects such as facial pruritus, nausea, or sedation. 8,29More recently, an alternative pharmacodynamic explanation suggested that it is the opioid effect and not the drug itself that spreads rostrally, thereby triggering various supraspinal effects. 36,37Modulating opioid action at the spinal cord causing differential activity of various ascending pathways has been a suggested mechanism for facial pruritus. 36Spinal opioids blocking ascending pathways that are tonically inhibiting supraspinal antinociceptive mechanisms has been suggested as a mechanism to explain supraspinal antinociceptive action. 37 

Our study does not allow differentiation between the pharmacokinetic and the pharmacodynamic model. However, a dose–response relationship seems evident for the rostral spread of analgesic effects, making the pharmacokinetic explanation appealing. 9,12Clearly, more research is necessary to evaluate the relative importance of pharmacokinetic and pharmacodynamic mechanisms that result in supraspinal opioid action after epidural administration.

The time course observed for the incidence of pruritus and nausea is in agreement with previous reports. 8,38,39Pruritus preceded and outlasted the occurrence of nausea. The time courses of pruritus and measured supraspinal analgesia were similar. This suggests that smaller doses of epidural morphine are necessary for causing supraspinal analgesic effects on heat pain than for causing nausea.

Compared with baseline, the heat and electrical pain tolerance typically were lower 2, 5, and 10 h after epidural injection of saline placebo but returned close to baseline after 24 h. This observation was not made when measuring the pain threshold. Although our study is not conclusive, several hypotheses about the mechanism underlying this observation can be formulated. First, injection of epidural saline may have caused hyperalgesia to strong nociceptive stimuli. However, it is more likely that injection of epidural saline has a local anesthetic effect if it has an effect at all. 40Second, repetitive administration of strong nociceptive stimuli may have caused hyperalgesia. However, such a phenomenon was not observed by other investigators who repetitively assessed the heat and electrical pain tolerance. 41Third, observed within-day fluctuation of the pain tolerance may be caused by a circadian change in pain sensitivity that does not become apparent when measuring the pain threshold. A study investigating the circadian sensitivity of experimentally induced headache found significant fluctuations for intense but not mild pain. 42Animal studies suggest that the sensitivity to nociceptive stimuli is lowest at the end of a resting period (dark cycle) and highest at the end of an activity period (light cycle). 43This matches the time course of observed changes in heat and electrical pain tolerance and suggests that the time-dependent variation is not caused by psychologic factors only.

In summary, this study reports long-lasting supraspinal analgesia to heat pain after a lumbar epidural dose of 5 mg morphine. Taken together with data from previous studies, it seems that morphine-mediated analgesic effects on heat pain spread more rostrally than analgesic effects on pricking or electrical pain. In a clinical context, this finding implies that epidural morphine may attenuate various types of pain up to a different dermatomal level.

The authors thank Lawrence J. Saidman, M.D., and Bruce MacIver, Ph.D., for helpful comments on the manuscript, and Byron W. Brown, Ph.D., for his statistical advice.