From the Departments of Anesthesia and Critical Care, Beth Israel Deaconess Medical Center (Drs. Jakubowski, Levy, Goor-Aryeh, Bajwa, Burstein and Ms. Collins); Department of Neurobiology and the Program in Neuroscience, Harvard Medical School (Dr. Burstein), Boston, Massachusetts.
Address all correspondence to Dr. Rami Burstein, Department of Anesthesia and Critical Care, Harvard Institutes of Medicine, Room 830, 77 Avenue Louis Pasteur, Boston, MA 02115.
Objective.—To determine whether delayed infusion of COX1/COX2 inhibitors (ketorolac, indomethacin) will stop migraine in allodynic patients, and suppress ongoing sensitization in central trigeminovascular neurons in the rat.
Background.—The majority of migraineurs seeking secondary or tertiary medical care develop cutaneous allodynia during the course of migraine, a sensory abnormality mediated by sensitization of central trigeminovascular neurons in the spinal trigeminal nucleus. Triptan therapy can render allodynic migraineurs pain free within a narrow window of time (20 to 120 minutes) that opens with the onset of pain and closes with the establishment of central sensitization. Can drugs that tackle ongoing central sensitization render allodynic migraineurs pain free after the window for triptan therapy has expired?
Methods.—Patients exhibiting migraine with allodynia were divided in two groups (n = 14, each): group 1 received delayed sumatriptan injection (6 mg) 4 hours after onset of attack—which failed to render them pain free—and ketorolac infusion (two 15-mg boluses) 2 hours later; group 2 received delayed ketorolac monotherapy 4 hours after onset of attack. Pain intensity (visual analog scale) and skin sensitivity (quantitative sensory testing) were measured when the patients were migraine free (baseline); 4 hours after onset of migraine (just before treatment); 2 hours after sumatriptan; 1 hour after ketorolac. In the rat, we tested whether infusion of ketorolac (0.4 mg/kg) or indomethacin (1 mg/kg) will block ongoing sensitization in peripheral and central trigeminovascular neurons. The induction of sensitization (using topical application of inflammatory soup on the dura) and its suppression by COX1/COX2 inhibitors were assessed by monitoring changes in spontaneous activity and responses to mechanical and thermal stimuli.
Results.—Patients had normal skin sensitivity in the absence of migraine, and presented cutaneous allodynia 4 hours after onset of migraine. In group 1, all patients continued to exhibit allodynia 2 hours after sumatriptan treatment, and none of them became pain free. However, 71% and 64% of the patients in groups 1 and 2, respectively, were rendered free of pain and allodynia within 60 minutes of ketorolac infusion. Nonresponders from both groups, in contrast to the responders, had had a history of opioid treatment. In the rat, infusion of COX1/COX2 inhibitors blocked sensitization in meningeal nociceptors and suppressed ongoing sensitization in spinal trigeminovascular neurons. This inhibitory action was reflected by normalization of neuronal firing rate and attenuation of neuronal responsiveness to mechanical stimulation of the dura, as well as mechanical and thermal stimulation of the skin.
Conclusions.—The termination of migraine with ongoing allodynia using COX1/COX2 inhibitors is achieved through the suppression of central sensitization. Although parenteral administration of COX1/COX2 inhibitors is impractical as routine migraine therapy, it should be the rescue therapy of choice for patients seeking emergency care for migraine. These patients should never be treated with opioids, particularly if they had no prior opioid exposure.
Electrophysiological studies on central trigeminovascular neurons in the spinal trigeminal nucleus and clinical studies on migraine patients have led us to suggest that the majority of patients seeking secondary and tertiary medical help have migraine that is associated with the phenomenon of central sensitization. We found that central sensitization represents a serious obstacle to the therapeutic action of triptans—a family of 5HT1B/1D receptor agonists most commonly prescribed for migraine treatment. In the present study, we tackled the problem of central sensitization using drugs that are commonly used as rescue therapy for patients who have already failed to respond to other treatments.
In the rat, we found that nociceptive neurons in the spinal trigeminal nucleus that receive convergent input from the dura and the periorbital skin can be rendered sensitized for several hours by briefly exposing the dura to a mixture of inflammatory agents.1 Using quantitative stimulation of the periorbital skin, we found that the sensitized neurons became super reactive to thermal (heat, cold) and mechanical stimuli.1,2 Consequently, we found that the majority (79%) of episodic migraine patients arriving at our pain clinic exhibit similar signs of periorbital allodynia (abnormal skin sensitivity) during the course of the attack.3 The likelihood of rendering allodynic patients pain free with triptan therapy diminished with the advance of the migraine attack, in contrast to nonallodynic patients who remained fully responsive to triptans throughout the attack.4 Going back to the rat model, we found that triptan intervention was inhibitory to the central trigeminovascular neurons at the onset of sensitization, but not after 2 hours of ongoing sensitization.5 Another important finding was that triptans produced no measurable inhibition over sensitized meningeal nociceptors, namely, the peripheral trigeminovascular units that convey nociceptive signals from the dura to the spinal trigeminal nucleus.6 We proposed that triptans exert their antimigraine action presynaptically in the dorsal horn by blocking signal transmission from axon terminals of peripheral trigeminovascular neurons to the cell bodies of their central counterparts.6
During a variety of pain conditions, the sensitization of dorsal-horn neurons begins with an activity-dependent phase (activity driven by incoming signals from the periphery) and shifts later to an activity-independent phase (activity that is self-sustained). The late phase, which in the case of migraine corresponds to phase where central sensitization and cutaneous allodynia are fully established, involves increased levels of cyclooxygenases in the spinal cord. This has led us to test whether ketorolac, a centrally acting cyclooxygenase (COX1/COX2) inhibitor7,8 commonly used as rescue therapy for patients who have already failed to respond to other treatments9–17 will (1) terminate migraine with allodynia in patients who already failed to respond to prior triptan treatment and (2) suppress ongoing sensitization in central trigeminovascular neurons in the rat.
Patient Selection.— This study was carried out in accordance with the ethical standards of the Committee on Clinical Investigation on Human Experimentation at Beth Israel Deaconess Medical Center and in accordance with the Helsinki Declaration of 1975, as revised in 1983. Patients included in the study were 17 to 55 years old who met the criteria of the International Headache Classification Committee for migraine,18 had 1 to 8 migraine attacks each month, and were able to give an informed consent. Excluded from this study were patients with chronic daily headache, peripheral nervous system injuries, and those using opioids for reasons other than migraine.
Experimental Protocol.— Patients exhibiting migraine with allodynia were divided in two groups (n = 14, each): group 1 received delayed sumatriptan injection (6 mg) 4 hours after onset of attack—which failed to render them pain free—and ketorolac infusion (two successive 15-mg boluses) 2 hours later; group 2 received delayed ketorolac monotherapy 4 hours after onset of attack. Pain intensity (visual analog scale) and skin sensitivity (quantitative sensory testing) were measured when the patients were migraine free (baseline); 4 hours after onset of migraine (just before treatment); 2 hours after sumatriptan; 1 hour after ketorolac.
Skin sensitivity was determined by measuring pain thresholds to cold, heat, and mechanical stimuli at the periorbital site of the referred pain, as described elsewhere.4,19 In healthy subjects, skin thresholds to pain are normally 42°C to 47°C for heat,20 5°C to 18°C for cold,20,21 and 75 to 281 g for mechanical stimulation.22 Therefore, we identified allodynia when skin threshold to pain was between 32°C and 40°C for heat, between 32°C and 20°C for cold, and <30 g for skin indentation.3 Meeting the criteria for any one modality was sufficient to determine that the patient was allodynic at the time of treatment.
Surgical Preparation.— All experiments were approved by the standing committee on animals of Harvard Medical School. Male Sprague-Dawley rats weighing 250 to 400 g were anesthetized with urethane (1.2 g/kg, i.p.) and fitted with an intratracheal metal tube to allow artificial ventilation, and a cannula inserted into the femoral vein for later infusion of drugs. Rats were placed in a stereotaxic apparatus, and core temperature kept at 37°C using a heating blanket. End-tidal CO2 was continuously monitored and kept within physiological range (3.5 to 4.5 pCO2). To apply chemical and mechanical stimulation onto the dura mater, we carefully removed the bone above the dorsal surface of the left hemisphere and transverse sinus and kept the overlying dura moist using a modified synthetic interstitial fluid (135 mM NaCl, 5 mM KCl, 1 mM MgCl2, 5 mM CaCl2, 10 mM glucose, and 10 mM Hepes, pH 7.2). To record the activity of meningeal nociceptors, we performed a second, more rostral craniotomy and lowered a platinum-coated tungsten microelectrode (impedance 500 kΩ) into the left trigeminal ganglion. To record the activity of central trigeminovascular neurons, we exposed the area between the obex and C2, which was kept moist with mineral oil, and lowered a platinum-coated tungsten microelectrode (impedance 4 to 6 MΩ) into the medullary dorsal horn. Anatomical localizations of central recording sites were identified postmortem as described elsewhere.1
Meningeal Nociceptors Recording.— Neurons in the trigeminal ganglion that exhibited constant latency responses to electrical stimulation of the dura were identified as meningeal nociceptors. Action potentials were processed using a real-time waveform discriminator (Spike 2, Cambridge Electronic Design, Cambridge, UK). Mechanical stimulation of the dura was applied using a servo, force-controlled mechanical stimulator (Aurora Scientific, Ontario, Canada) with a flat-ended cylindrical plastic probe, 0.5 mm in diameter.23 Each trial was 4-minutes long and consisted of ramp-and-hold stimulations at four incremental forces (rise time 100 msec, stimulus width 2 seconds, interstimulus interval 60 seconds). Stimuli that evoked 1 to 2 spikes/sec above baseline were considered as threshold. Trials were delivered every 10 minutes, and spontaneous firing was recorded between trials.
Central Trigeminovascular Neurons Recording.— Trigeminovascular neurons in the spinal trigeminal nucleus were first identified based on their responses to electrical stimulation of the dura. Neuron exhibiting Aδ- and C-fiber volleys were characterized for their baseline responses to mechanical indentation of the dura (using calibrated von Frey hairs, range: 0.06 to 3.6 g) and to mechanical and thermal stimulation of the skin as described before.1 Stimuli were repeated at least three times at 30-minute intervals and only neurons showing less than 10% variability in responses were included in the study.
Induction of Sensitization.— Topical application (for 5 minutes) of an inflammatory soup (IS) containing a mixture of inflammatory mediators (histamine, serotonin, bradykinin, all at 1 mM, and 0.1 mM prostaglandin E2, pH 5.5)24–26 was used to induce activation and sensitization as described earlier.1,27 The establishment of peripheral and central sensitization was verified 2 hours after stimulation of the dura with IS using the following criteria: (1) increased neuronal firing rate, (2) increase in magnitude of responses to mechanical stimulation of the dura and skin, (3) expansion of dural and cutaneous receptive fields.
COX1/COX2 Inhibition.— Two hours after induction of sensitization, intravenous bolus of ketorolac (0.4 mg/kg) was administered in the study of meningeal nociceptors and central trigeminovascular neurons. An additional set of central neurons was studied using indomethacin (1 mg/kg). Ongoing activity of meningeal nociceptors and their responsiveness to mechanical stimulation of the dura were monitored every 10 minutes over 1 hour after ketorolac administration. Ongoing activity of central trigeminovascular neurons and their responsiveness to mechanical and thermal stimulation of the skin were measured 1 hour after indomethacin administration.
Statistical Analysis.— Data analysis was performed using nonparametric statistics.28,29 Pain levels in migraine patients were analyzed using Friedman two-way analysis of variance. Post hoc paired comparisons were performed using one-tailed Wilcoxon matched-pairs signed-ranks test. Level of significance was set at 0.05.
Migraine Pain.— The 28 patients that participated in the study had a history of migraine with allodynia that proved unstoppable with delayed triptan treatment. Of these, 19 patients (10 of group 1 and 9 of group 2) were rendered pain free within 1 hour of delayed ketorolac infusion (responders, Figure 1A,B). The remaining 9 patients (4 of group 1 and 5 of group 2) continued to sustain moderate to severe headache for at least 2 hours after ketorolac treatment (nonresponders, Figure 1C,D). Pretreatment pain intensity (at 4 hours after onset of migraine) was identical across responders and nonresponders from groups 1 and 2 (Figure 1A-D). Analyses of demographic data (eg, age, age of onset, frequency and duration of attacks) and migraine symptoms (eg, aura, nausea, vomiting, phobias) showed no significant difference between responders and nonresponders. The only apparent difference pointed to opioid history: 9/9 nonresponders, and only 1/19 responders, were habituated to treating their migraine with opioids such as meperidine (demerol), butophanol (stadol), oxycodone (percocet), and hydrocodone (codeine).
Cutaneous Allodynia.— Skin sensitivity was normal in all patients in the absence of migraine. Four hours after the onset of migraine (before any treatment), 60% of the patients were exhibiting allodynia for three modalities, the remaining 40% were exhibiting allodynia for two modalities (heat, cold, and mechanical allodynia were presented, respectively, by 93%, 74%, 85% of the patients). Patients that received sumatriptan injection continued to exhibit the same levels of allodynia 2 hours later in association with their ongoing headache (Figures 2A,C and 3A,C). Patients that were rendered pain free within 1 hour of ketorolac infusion (responders) were also rendered allodynia free, ie, their pain thresholds to heat, cold, and mechanical stimulation of the periorbital skin returned to normal (Figure 2A-D). Patients that continued to sustain migraine headache 2 hours after ketorolac infusion (nonresponders) remained allodynic, regardless of whether (Figure 3A,C) or not (Figure 3B,D) they receive sumatriptan 2 hours earlier.
Animal Study.— The observation that ketorolac infusion aborted migraine after the establishment of allodynia sent us to the bench to study the effects of cyclooxygenase inhibitors on sensitized peripheral and central trigeminovascular neurons in the rat.
Peripheral Sensitization.— Preliminary data from three sensitized meningeal nociceptors suggest that infusion of ketorolac normalized their ongoing activity and their sensitivity to mechanical stimulation of the dura. An example of one such neuron is illustrated in Figure 4. Sensitization was induced by stimulating the dura with IS for 5 minutes, resulting in long-lasting decrease in threshold and increase in magnitude of its responses to quantitative mechanical stimulation of the dura (compare baseline to 60, 120, 180 minutes). Within 40 minutes of ketorolac infusion, the changes induced by IS were reversed, and response threshold and magnitude returned to baseline values.
Central Sensitization.— Intravenous infusion of two COX1/COX2 inhibitors effectively reversed central sensitization previously induced by IS application to the dura.
Data from three sensitized trigeminovascular neurons showed that skin sensitivity was normalized with the COX1/COX2 inhibitor indomethacin (Figure 5). Before indomethacin, neuronal response magnitude to mechanical stimulation of the periorbital skin (brush, pressure, pinch) increased and facial receptive field expanded after stimulating the dura with IS (Figure 5, cf. green vs. red). Within 90 minutes of indomethacin infusion the changes induced by IS were reversed, and response magnitudes returned to baseline values and the receptive field shrunk (Figure 5, cf. red vs. blue).
Data from six sensitized trigeminovascular neurons in the spinal trigeminal nucleus showed that infusion of ketorolac normalized ongoing activity (Figure 6), sensitivity to mechanical stimulation of the dura (Figure 7), and sensitivity to mechanical and thermal (heat and cold) stimulation of the periorbital skin (Figure 8). Before ketorolac, neuronal ongoing activity and response magnitudes to stimulation of the dura and skin increased after stimulating the dura with IS (Figures 6–8, cf. green vs. red). Within 90 minutes of ketorolac infusion, neuronal ongoing activity and response magnitudes to stimulation of the dura and skin returned to baseline values (Figures 6–8, cf. red vs. blue).
We have previously shown that patients who develop periorbital allodynia during migraine can be rendered pain free using triptan therapy soon after the onset of the attack, but not hours later. The present study shows that ketorolac, a COX1/COX2 inhibitor, provides a rescue therapy for allodynic patients who already missed the critical period for triptan therapy. Belated infusion of ketorolac terminated both the headache and the allodynia in 68% of 28 patients (responders) within 60 minutes of treatment, independent of whether or not ketorolac infusion was preceded by sumatriptan injection. In the remaining patients (nonresponders), ketorolac infusion did not terminate the headache, and it did not abolish allodynia. Since periorbital allodynia is mediated by sensitization of trigeminovascular neurons in the spinal trigeminal nucleus, we conclude that ketorolac infusion terminated migraine with ongoing allodynia by suppressing central sensitization.
We believe that migraine with ongoing allodynia is mediated by an activity-independent form of central sensitization in the spinal trigeminal nucleus. This form of sensitization is initiated by prolonged barrage of incoming signals from nociceptors30 that activate nociceptive neurons and glial cells in the spinal cord.31 The release of prostaglandins from the activated neurons in the dorsal horn facilitates prostaglandin release from the activated glia, which, in turn, promotes the hyperexcitability of adjacent nociceptive neurons.32,33 Therefore, inhibition of spinal cyclooxygenase activity, which is the rate-limiting enzyme in prostaglandin biosynthesis, may be the mechanism by which parenteral ketorolac treatment terminates migraine associated with ongoing allodynia and central sensitization.
We were surprised to find that the nonresponders, in contrast to the responders, had a long history of opioid therapy. The mechanisms by which opioids may modify properties of central nociceptive neurons and render them unresponsive to other analgesic drugs is largely unknown. Prolonged exposure to opioids may result in tonic activation of descending pain facilitatory pathways arising from the rostral ventromedial medulla,34 and increased expression of dynorphin in the spinal cord.35 In contrast to its action as a kappa-opioid agonist, dynorphin has significant extraopioid activities that enhance nociception,36–38 potentially by promoting release of excitatory amino acids,39 substance P,40 and CGRP41 from primary afferent nociceptors.
The finding that nonresponders had a long history of opioid therapy should raise a red flag regarding the practice of treating migraineurs with opioids when they seek help in the emergency department (ED). Our patients testified that they were introduced to opioids in the ED, and came to believe this was the most effective means of terminating their migraine after having failed to respond to over-the-counter COX1/COX2 inhibitors, or prescription drugs such as triptans. Using data of a 1998 Nationwide Hospital Ambulatory Medical Care Survey of 811,419 adult migraineurs visiting the ED, Vinson17 has pointed out that 411,350 (51%) were treated with opioids (meperidine, nalbuphine, butophanol, morphine) and, even more alarming, that 77% of those did not receive any nonopioid abortive medication prior to the opioid therapy. We believe it is imperative that migraine patients arriving in the ED for help, particularly those with no prior exposure to opioids, be treated with parenteral COX1/COX2 inhibitors, not with opioids.
Compared to our previous studies in the rat,5 it appears that COX1/COX2 inhibitors differ from triptans in terms of their effects on peripheral and central trigeminovascular neurons. Triptans appear to act in the dorsal horn to block signal transmission from meningeal nociceptors to the central trigeminovascular neurons, whereas COX1/COX2 inhibitors appear to silence the peripheral and central neurons themselves. In spite of these advantages, it should be emphasized that the bioavailability of COX1/COX2 inhibitors to neurons in the spinal cord depends on the route of drug delivery.42 Indeed, the same allodynic patients in our study that were rendered pain free by ketorolac infusion testified that oral formulations of COX1/COX2 inhibitors were ineffective in aborting their migraine. Since parenteral drug administration is impractical as routine migraine therapy, it will be useful to develop oral formulations of COX1/COX2 inhibitors with comparable bioavailability to the spinal cord.
Acknowledgments: The study was supported by NIH grants DE13347, NS051484, and NS35611 to Dr. Burstein.