IN homeothermic species, a thermoregulatory system coordinates defenses against cold and heat to maintain internal body temperature within a narrow range, thus optimizing normal physiologic and metabolic function. The combination of anesthetic-induced thermoregulatory impairment and exposure to a cool environment makes most unwarmed surgical patients hypothermic. 1–7Although shivering is but one consequence of perioperative hypothermia, and rarely the most serious, it occurs frequently (i.e. , 40–60% after volatile anesthetics), 8,9and it remains poorly understood. While cold-induced thermoregulatory shivering remains an obvious etiology, the phenomenon has also been attributed to numerous other causes.
Our first goal is to review the organization of the thermoregulatory system, and particularly the physiology of postanesthetic shivering. We then discuss the pharmacology of thermoregulation and review the putative mechanisms and sites of action of various antishivering drugs.
Neuronal Networks Controlling Thermoregulation
Historically, the lateral spinothalamic tract was considered the sole thermoafferent pathway, projecting to the hypothalamic thermoregulatory centers. 10However, evidence suggests that the majority of these ascending pathways terminate in the reticular formation 11,12and that thermosensitive neurons exist at several regions outside the preoptic-anterior hypothalamus, including the ventromedial hypothalamus, 13the midbrain, 14–17the medulla oblongata, 18,19and the spinal cord. 20,21Multiple inputs from various thermosensitive sites are integrated at numerous levels within the spinal cord and brain to provide a coordinated pattern of defense responses. 22,23
The temperature-regulating system of mammals is often divided into three components: thermosensors and afferent neural pathways, integration of thermal inputs, and effector pathways for autonomic and behavioral regulation. The major afferent thermoregulatory structures and the efferent shivering pathway are depicted in figure 1.
Thermosensors and Afferent Neural Pathways
The thermosensitivity of the spinal cord and its thermoregulatory significance is beyond doubt 24–29and has been reviewed comprehensively. 20Its ability to sense and modulate thermal signals was pivotal for development of the currently accepted multiple-input, multilevel concept of thermoregulation. 22In fact, all thermoregulatory effector mechanisms are modulated by spinal cord temperature. In intact 30and chronically spinalized 31dogs and rabbits, 32selective cooling of the spinal cord induces cold tremor. In humans, shivering is rare and of low intensity below the level of injury in patients with spinal cord transection. 33
The Extrahypothalamic Brain Stem.
Thermosensitive sites that are not associated with defined anatomic structures appear to be dispersed in the lower brain stem. 34Experiments in rats suggest that heat gain responses are powerfully regulated by a tonic inhibitory mechanism located in the midbrain and upper pons. 35In the reticular formation ‡of the rat, two anatomically separate groups of neurons are involved in thermal responsiveness and control of thermoregulatory muscle tone and shivering. 14A comparative study in vertebrates also concluded that peripheral thermal input to the hypothalamic areas is via the polysynaptic nonspecific reticular areas in the brainstem. 12
The Nucleus Raphe Magnus and the Subcoeruleus Area.
The nucleus raphe magnus in the medulla contains a relatively high percentage of serotonergic thermoresponsive neurons, with a preponderance of warm responsive neurons. 18The locus subcoeruleus is a circumscribed area in the pons ventromedially to the locus coeruleus, 36which contains the largest cluster of noradrenergic neurons in the brain. 37The nucleus raphe magnus and the subcoeruleus area appear to be important relay stations in the transmission of thermal information from skin to hypothalamus. 36,38These areas seem to be responsible for the modulation rather than the generation of thermal afferent information. 39–41
Integration of Thermal Inputs
Most investigators accept that the preoptic region of the anterior hypothalamus is the dominant autonomic thermoregulatory controller in mammals. However, preoptic-anterior hypothalamus neurons also respond to nonthermal information, e.g. , reproductive hormones, 42plasma osmolality, 43–45glucose concentration, 43,46blood pressure, 47noxious stimuli, 48carbon dioxide, 49and emotional stimuli. 50Much of the excitatory input to warm-sensitive neurons in the preoptic-anterior hypothalamus comes from the hippocampus, 51which links the limbic system (emotion, memory, and behavior) to thermoregulatory responses.
In addition, the level of activity in preoptic neurons is modulated by arousal state 52and suprachiasmatic nucleus activity, 53which may explain why changes in body temperature are associated with sleep and circadian rhythms. Thus, warm-sensitive neurons in the preoptic-anterior hypothalamus not only sense core temperature but also compare local information with thermal and nonthermal synaptic afferents arriving over ascending pathways. These interactions are inevitable because the thermoregulatory system has few specific effector organs and must be understood as a part of the adaptive responses of the organism as a whole. 54
Classic neuronal models of the hypothalamus functionally separate the integrative and effector neurons controlling thermoregulatory responses. 55However, electrophysiologic studies suggest that some anterior hypothalamic neurons act as sensors as well as integrators 56and suggest a link between neuronal firing rate and the range of thermosensitivity. 57The model of Boulant 57identifies different groups of warm-sensitive neurons, distinguished by their spontaneous firing rates. Varying combinations of afferent inputs trigger different groups of warm-sensitive neurons, and effector mechanisms are therefore activated in an orderly fashion (fig. 2).
All neuronal models of temperature regulation use the concept of the common central command: multiple inputs are integrated into a common efferent signal to the effector systems. 58In both animals 59and humans, 60effector mechanisms are called on in an orderly fashion, ensuring optimal regulation at minimum cost. The principal defenses against hypothermia in humans include skin vasomotor activity, nonshivering thermogenesis, shivering, and sweating.
Heat loss is normally regulated without the major responses of sweating or shivering because cutaneous vasodilation and vasoconstriction usually suffice. 56,61Thermoregulatory vasoconstriction 62decreases cutaneous heat loss 63,64and constrains metabolic heat to the core thermal compartment. 65,66This usually prevents body temperature from decreasing the required additional 1°C required to activate intraoperative shivering. 67–70Normal thermoregulatory shivering is thus a last-resort defense that is activated only when behavioral compensations and maximal arteriovenous shunt vasoconstriction are insufficient to maintain core temperature.
Nonshivering thermogenesis is the result of cellular metabolic processes that do not produce mechanical work. Thermoregulatory nonshivering thermogenesis has been demonstrated in the human neonate 71and in rodents, but its existence in adult humans is uncertain, 72as it is not observed in anesthetized adults 73or infants. 74
Shivering is an involuntary, oscillatory muscular activity that augments metabolic heat production. Vigorous shivering increases metabolic heat production up to 600% above basal level. 75However, a doubling of metabolic heat production is all that can be sustained over long periods. 76The fundamental tremor frequency on the electromyogram in humans is typically near 200 Hz. This basal frequency is modulated by a slow, 4–8 cycles/min, waxing-and-waning pattern (fig. 3). 77,78
Shivering is elicited when the preoptic region of the hypothalamus is cooled. 79Efferent signals mediating shivering descend in the medial forebrain bundle. 80Classically, a central descending shivering pathway was thought to arise from the posterior hypothalamus. 81–85Although the preoptic-anterior hypothalamus is thought to suppress shivering by inhibition of the posterior hypothalamus, 86experimental evidence is lacking. Thermally induced changes in neuronal activity in the mesencephalic reticular formation 87and the dorsolateral pontine and medullary reticular formation 14exert descending influences on the spinal cord that increase muscle tone. 14It remains to be determined whether the reticulospinal neurons receive synaptic input directly from the preoptic-anterior hypothalamus or from the posterior hypothalamus.
Spinal α motor neurons and their axons are the final common path for both coordinated movement and shivering. 88A typical cold tremor has a specific rhythm in the form of grouped discharges in the electromyography. 89–91One hypothesis suggests that excitability of motor neurons is inversely proportional to cell size. 92,93During continued cold stimulation of the skin or the spinal cord, motor neurons are recruited in sequence of increasing size, starting with the small γ motor neurons that are followed by the small tonic α motor neurons, and finally, the larger phasic α motor neurons. 92,94,95
The larger α motor neurons are more likely to manifest synchronized discharges than smaller ones. 96Synchronization of motor neurons during shivering may be mediated by recurrent inhibition through Renshaw cells, a group of inhibitory interneurons identified in the cat. 97,98Reflex activation of α motor neurons via the γ muscle spindle loop (instability of the stretch reflex feedback system), is another potential but controversial mechanism that could determine the rhythm and frequency of α motor neurons discharges. 99–101
Postanesthetic Shivering and Shivering-like Tremor
Patients report that shivering is remarkably uncomfortable, and some even find the accompanying cold sensation worse than surgical pain. Moreover, shivering per se may aggravate postoperative pain simply by stretching surgical incisions. Shivering also occasionally impedes monitoring techniques, 102,103increases intraocular 104and intracranial 105pressures, and is especially disturbing to mothers during labor and delivery. 106
Shivering can double or even triple oxygen consumption and carbon dioxide production, although the increases are typically much smaller. 107,108These large increases in metabolic requirement might predispose to difficulties patients with existing intrapulmonary shunts, fixed cardiac output, or limited respiratory reserve. However, shivering is rare in elderly patients 109–111because age per se impairs normal thermoregulatory control. 112–117Because shivering intensity is markedly reduced in elderly and frail patients, it is unlikely that shivering itself provokes serious adverse outcomes in these patients.
Likewise, shivering is rarely associated with clinically important hypoxemia because hypoxia itself inhibits this response. 118,119Morbid cardiac outcomes associated with mild perioperative hypothermia appear to be mediated by a mechanism more subtle than shivering— perhaps the associated marked increase in plasma catecholamine concentrations. 120
Abnormal Tremor Patterns
Shivering is common in hypothermic patients recovering from general anesthesia. 121–123The conventional explanation for postanesthetic tremor is that anesthetic-induced thermoregulatory inhibition abruptly dissipates, thus increasing the shivering threshold toward normal. Discrepancy between the persistent low body temperature and the now, near-normal, threshold activates simple thermoregulatory shivering. Difficulties with this proposed explanation include the observations that tremor frequently is not observed in markedly hypothermic patients 122and that tremor occurs commonly in normothermic patients. 124However, a subsequent study 78suggested that special factors related to surgery (such as stress or pain) might contribute to the genesis of postoperative tremor because it failed to identify any shivering-like activity in normothermic volunteers. Pain might facilitate shivering-like tremor in both postoperative patients 125and in women having spontaneous term labor. 126
Any increase in the thermoregulatory set-point (fever) may be associated with normal thermoregulatory shivering in normothermic or even hyperthermic patients. 124,127Surgical stress may increase the thermoregulatory set-point in the postoperative period: even in the absence of clinically evident signs of infection, 25% of postoperative patients reach core temperatures of 38°C, and 50% of them reach 38.4°C. 128Of course, there are many other reasons surgical patients might develop a fever, including infection, atelectasis, and release of pyrogenic substances by injured tissues.
Three patterns of muscular activity were observed in hypothermic volunteers during emergence from isoflurane anesthesia. 78The first was a tonic stiffening and appeared to be largely a direct, non–temperature-dependent effect of isoflurane anesthesia. Near 0.3% end-tidal isoflurane concentration, a second pattern was overt: synchronous, tonic waxing and waning. This was by far the most common pattern and resembled that produced by cold-induced shivering in unanesthetized volunteers, or “true” thermoregulatory shivering. 77The third observed pattern was a spontaneous electromyographic clonus that required both hypothermia and residual isoflurane end-tidal concentrations between 0.4 and 0.2% (fig. 4). During epidural anesthesia, synchronous waxing-and-waning patterns were present; however, no abnormal (i.e. , clonic) electromyogram patterns were detected. 129
Despite alternative etiologies in some patients, normal thermoregulatory shivering in response to core and skin hypothermia remains by far the most common cause of postoperative shivering. The remainder of this review therefore focuses on normal thermoregulatory shivering.
Dependence of Thermosensitivity on the State of Arousal
Although it is not possible to focus a thermal stimulus to a single cell, there are thermosensitive units in the hypothalamus that might be considered thermoresponsive. 130These units may be activated by direct thermal stimulation or by other interconnected interneurons responding to thermal stimulation of the skin or distant areas in the central nervous system. Thermoresponsiveness of these units is not constant but varies significantly over time 52,131and depends on the state of vigilance 52,132and cortical activity. 133–135
Recent work demonstrated the potential for arousal state to combine with thermal influences to create the appearance that cells are thermosensitive or thermoresponsive when, in fact, they may not be responding directly to temperature. Thus, when electroencephalographic state changes are taken into account, all changes in firing rate of preoptic-anterior hypothalamic cells that appeared to be responsive to changes in skin temperature are associated with electroencephalographic state changes. 136Single-unit responses in the rostral ventromedial medulla, which consists of the nucleus raphe magnus and adjacent brain stem regions, are not specific for temperature manipulations but reflect changes in electroencephalogram–electromyogram activity, which in turn is determined by a variety of factors, including thermal and noxious stimuli. 137Similar results (no thermoresponses observed within a given electroencephalographic state) were obtained for single-unit activity in the subcoeruleus area. 138
Pharmacologic Modulation of Shivering
Several classes of substances, including biogenic monoamines, cholinomimetics, cations, endogenous peptides, and possibly N -methyl-d-aspartate (NMDA) receptor antagonists, appear to modulate central thermoregulatory control mechanisms. In this section, we discuss these chemically induced changes in thermosensitivity and modulation of thermosensitivity by drugs used to control postanesthetic shivering. The predominant site of action of the discussed drugs is in most, if not all, instances difficult to establish.
The Monoamine Theory of thermoregulation was born with Feldberg and Myers’ suggestion in 1963 that the balance of norepinephrine and serotonin (5 hydroxytryptamine [5-HT]) in the preoptic-anterior hypothalamus controls the body temperature set-point. 157Initially, specific thermoregulatory responses were demonstrated in the cat by direct intracerebroventricular injection of adrenergic and serotonergic neurotransmitters. The monoamines seemed to have opposite effects: 5-HT caused shivering and vasoconstriction and a concomitant increase in core temperature, whereas norepinephrine and epinephrine lowered the normal resting temperature of the cat and attenuated the 5-HT-induced hyperthermia. 158
In similar experiments, other species reacted in the opposite way, i.e. , norepinephrine increased and 5-HT decreased body temperature. These interspecies differences have been reviewed in detail by other investigators. 159–161Contradictory results were reported for monoamines in a given species as well, and were attributed to differences in dosage, 162microinjection technique, 163ambient temperature, 164,165and other factors. 166
Neurotransmitters modulate the synaptic input on temperature-sensitive neurons and may have profound effects on their firing rates and range of thermosensitivity. The way thermal signals from cold and warm sensors are integrated in the hypothalamus led to speculation that the set-point of the thermoregulatory system could be easily manipulated if the few specific inputs consisted of certain transmitters. 167This turned out to be a considerable oversimplification because thermoregulatory thresholds are determined by multiple modulatory thermal and nonthermal inputs (that are not all monoaminergic) and take place at all levels of hierarchy in the thermoregulatory system. Nevertheless, the balance between the modulatory 5-HT and norepinephrine inputs may be responsible for short- and long-term thermoregulatory adaptive modifications of the shivering threshold. 39,55,168
Norepinephrine microdialyzed into the preoptic area of conscious guinea pigs reduces core temperature, a reduction that is abolished by coadministration of the α2-adrenoceptor antagonists yohimbine and rauwolscine. 169The α2-adrenoceptor agonist clonidine evokes dose-dependent reductions in core temperature, whereas α1-, β1-, and β2-adrenoceptor agonists and antagonists do not induce significant changes in core temperature. Elevation of the ambient temperature to 40°C induces a selective increase in the release of norepinephrine perfusates collected with a push-pull cannula from the rostral hypothalamus of the cat, 170whereas decreasing the ambient temperature to 2°C markedly reduces the norepinephrine release from the preoptic-anterior hypothalamic area of the rat. 171
5-Hydroxytryptamine may influence both heat production and heat loss pathways. Apart from interspecies differences, 5-HT elicits divergent thermoregulatory responses at different thermosensitive sites within the hypothalamus. Injection of 5-HT into the preoptic area of cats evokes hypothermia accompanied by vasodilation. 172When 5-HT is injected into the rostral hypothalamus of cats, hyperthermia associated with shivering is evoked. 172In rat midbrain slice preparations, the majority of warm-sensitive units and all cold-sensitive units are inhibited by 5-HT. 173In contrast, 5-HT activates the majority of temperature-sensitive units in the medulla oblongata of the rat. 173
Opposite modulatory inputs from noradrenergic and serotonergic neurons in the lower brain stem modify the composite skin temperature signal integrated at the level of the hypothalamus, thereby shifting the thresholds and slopes for thermoregulatory responses. 168In different physiologic situations, e.g. , during cold adaptation or during fever, the interthreshold range (temperatures between the sweating and shivering thresholds) widens or narrows. In cold-adapted guinea pigs, for example, serotonergic input dominates and produces a wide interthreshold zone with an average body temperature of 38°C (compared with 39°C when the norepinephrine input dominates). 174Similarly, the interthreshold zone nearly doubles in cold-adapted humans, mainly because the shivering threshold is reduced by approximately 1°C to 35.4°C. 175Despite multiple confounding factors, there is increasing evidence for the involvement of monoaminergic brain systems in adaptive changes in thermoregulation (fig. 5). 55,161
Dopamine injected into the hypothalamus of the unanesthetized monkey in the same range of doses as norepinephrine induced hypothermia, but to a lesser degree. 176In single-unit studies, the spontaneous firing rate of cold-sensitive neurons of the cat's hypothalamus decreased when dopamine was applied iontophoretically. 177Perfusion with dopamine increased the firing rate of many warm-sensitive neurons in hypothalamic slices. 178In a hot environment, dopamine was increased in push-pull perfusates within the preoptic-anterior hypothalamic area of the cat. 179During cold exposure, shivering thermogenesis is inhibited after intracerebroventricular injection of dopamine in the goat. 180Furthermore, dopamine in the nigrostriatal system may play a role in central thermoregulation. 181
Histaminergic pathways also may be involved in central thermoregulation, via both H1and H2histamine receptors, as demonstrated in behavioral studies. 182–184There is some evidence for histaminergic pathways in the rostral hypothalamus involved in thermoregulation and integration with other monoaminergic thermoregulatory pathways, as reviewed previously. 185
Nefopam, 186an analgesic with powerful antishivering properties, 105is a potent inhibitor of synaptosomal uptake of 5-HT, norepinephrine, and dopamine, 187and slightly lowers normal body temperature. 188Tramadol 189,190is an antishivering drug 146with a similar mechanism of action: it inhibits the reuptake of 5-HT, 191norepinephrine, 192and dopamine 193and facilitates 5-HT release. 191Despite different degrees of opioid-like characteristics in preclinical tests, 192tramadol lacks significant naloxone reversibility in humans. 194,195In human volunteers, a high dose of naloxone only partially reverses the antishivering effect of tramadol. 146Cerebral α2adrenoceptors are thought to play a role in the attenuation of postoperative shivering by tramadol. 196
α2-Adrenergic agonists 197hyperpolarize neurons, presumably by increasing potassium conductance through Gi-coupled proteins. 198,199This, in turn, suppresses neuronal firing, 200which is linked to the range of thermosensitivity. 57Furthermore, activation of α2-adrenoceptors suppresses N-type calcium entry into nerve cells, 200which depresses neurotransmitter release. 201A greater retention of Ca2+ions on the neuron's surface stabilizes the cell membrane and lowers the firing rate of heat gain units in the posterior hypothalamus. 202
The antihypertensive drug ketanserin also interferes with postanesthetic shivering; however, the efficacy of ketanserin is rather low. 141,203Ketanserin is an antagonist with high affinity for both 5-HT2receptors 204 205,206and α1adrenoceptors. 207,208Similar to other α1-adrenoceptor antagonists (e.g. , prazosin), ketanserin acts indirectly via facilitation of a central presynaptic α2-adrenoceptor mechanism in the lower brainstem. 208
5-Hydroxytryptamine type 3 receptor antagonists, known as antiemetic drugs, are currently under investigation for a possible role in the prevention and treatment of postanesthetic shivering. 154,209There are almost no animal data available about 5-HT3receptor–mediated temperature-regulating mechanisms. 210
Sites of Action.
The effects of nefopam 187and tramadol 191,192at the level of the pons may partially explain their antishivering effect. In the rat locus coeruleus, tramadol and its main metabolite, O-desmethyltramadol, reduce neuronal firing rate and hyperpolarize neurons in a concentration-dependent manner. 211The locus coeruleus appears to be a proshivering center that activates heat production in rodents. 55The locus coeruleus is also the main noradrenergic nucleus involved in the descending pain-control system, 212with its activity regulated by α2autoreceptors. §5-HT1Areceptors modulate responses mediated by α2Aadrenoceptors in the locus coeruleus. 213In humans, α2-adrenoceptor antagonism with yohimbine significantly reverses the analgesic effects of tramadol. 195
Racemic tramadol and its (+) enantiomer significantly reduce 5-HT uptake and increase stimulated 5-HT efflux in the dorsal raphe nucleus. 214The dorsal raphe nucleus is part of the brainstem raphe complex and is considered one of the most important nuclei in the modulation of pain in the central nervous system. 212The nucleus raphe magnus is an antishivering center that activates heat loss mechanisms and inhibits thermogenesis during cold adaptation. 55,2155-HT is the major neurotransmitter in the raphe nuclei, but half of the raphe cells that project to the spinal cord are not serotoninergic. 216There is also a significant amount of norepinephrine in the nucleus raphe magnus, and approximately 10% of nucleus raphe magnus serotoninergic cells express α2adrenoreceptors. 217
An inhibitory role of the nucleus raphe magnus on shivering is caused by projections to hypothalamic units and by a second pathway descending from the nucleus raphe magnus to the spinal cord where dorsal horn cells are inhibited presynaptically. 40Postsynaptic activation of noradrenergic units in the subcoeruleus region inhibit warm-responsive units in an area between the anterior hypothalamus and the posterior hypothalamus, and in the posterior hypothalamus itself. 38Other projections of the subcoeruleus region descend to the pons and medulla and to motor neurons and autonomic preganglionic cell groups in the spinal cord. 218Just as descending inhibition restricts transmission of pain signals, efferents to the dorsal horn of the spinal cord may inhibit cutaneous thermal input. 215,219,220However, this assumption remains controversial. 221In the very least, descending 5-HT terminals from the locus coeruleus make intimate contact with motor neurons, mostly via cord internuntials. 222–224However, the role of these nerve terminals in the modulation of shivering remains to be established.
An anatomic target for the antishivering effect of α2-adrenergic agonists can be found at three levels. First, a small intravenous dose of clonidine reduces the spontaneous firing rate in the locus coeruleus 225and indirectly reduces norepinephrine-induced firing of serotoninergic neurons in the dorsal raphe nucleus. 226Second, the action of α2-adrenergic agonists in the locus coeruleus may also increase activation of α2adrenoceptors in the spinal cord. 227Intrathecal α2-adrenergic agonists are known to release dynorphin (a κ-opioid agonist) 228and to stimulate norepinephrine and acetylcholine release. 229Dynorphin is present in high concentration in the spinal cord 230and is involved in antinociception. Norepinephrine and acetylcholine suppress responses of wide-dynamic-range neurons to noxious stimulation in the spinal dorsal horn. As postulated, depressor effects of these neurotransmitters at the dorsal horn may modulate cutaneous thermal input additional to noxious and mechanoreceptive transmission. 215Third, the hypothalamus contains a high density of α2adrenoceptors. Norepinephrine microdialyzed into the hypothalamus, for example, activates α2adrenoceptors, reduces metabolic heat production, and produces hypothermia. 169Pretreatment of the preoptic-anterior hypothalamic area with the selective α2-adrenoceptor agonist yohimbine inhibited the hypothermic response of clonidine. 231
In single-unit studies of the preoptic-anterior hypothalamus in cats, 232rats, 233and many other species, 161the effect of acetylcholine on thermosensitive neurons remains inconclusive. The muscarinic or nicotinic cholinoceptors may be involved, because both acetylcholine and nicotine apparently induce vasoconstriction, shivering, and a hyperthermic reaction when injected into the hypothalamus of a conscious monkey. 176,234Antimuscarinic drugs have been used to demonstrate the physiologic role of the central cholinergic system in thermoregulation. However, a lack of selectivity and other methodologic problems influenced the results: for example, intracerebroventricular administration of atropine in the rabbit suppresses shivering and causes hypothermia, 235whereas rats become hyperthermic when atropine is injected into the hypothalamus. 236In rabbits, intravenous injection of nicotine stops shivering. 237
There is evidence in monkeys that cholinergic activity in the hypothalamus modulates heat gain (shivering) during heat or cold stress. Release of acetylcholine, for example, is markedly increased by 88% at the active acetylcholine-releasing sites within the preoptic-anterior hypothalamic area by peripheral cooling, but suppressed by 80% at the same perfusion sites by peripheral warming. 238Within the posterior hypothalamus, cold stress doubles acetylcholine release. Injection of a large dose of a cholinomimetic into the posterior hypothalamus causes hypothermia, however, presumably because of a “depolarizing blockade” of the cholinergic receptor system involved in heat production. 161,176
In the brain stem, cholinoceptors also may participate in thermoregulation, interacting with monoaminergic and peptidergic systems. Microinjection of the cholinoceptor agonists, carbachol and pilocarpine, into the mesencephalic nucleus raphe magnus caused significant hyperthermia, which was blocked by local pretreatment with a muscarinic receptor antagonist as well as a nicotinic receptor antagonist. 239Intracerebroventricular pretreatment with a 5-HT reuptake blocker significantly inhibited the carbachol-induced hyperthermia, which suggests that the hyperthermia is caused by an inhibition of a 5-HT–sensitive hypothalamic heat loss mechanism. Injection of carbachol into the periaqueductal gray area of rat brain results in hypothermia, probably mediated by neurotensin. 240
Physostigmine is as effective in preventing postanesthetic shivering as meperidine and clonidine. 144Physostigmine is the classic centrally acting cholinesterase inhibitor but is relatively nonselective. The analgesic effect of physostigmine may be mediated via cerebral cholinergic muscarinic receptors, 241but serotoninergic receptors 242and an endorphinergic mechanism 243,244may also be involved. Analgesia after intrathecal administration of anticholinesterase is mediated through muscarinic receptors, and there is a synergistic interaction with intrathecal μ-opioid and α2-adreneric agonists. 245It is unknown if the same receptors also mediate the thermoregulatory effects of physostigmine.
In a prospective, randomized, double-blind study, healthy adult patients who were premedicated with an anticholinergic had a significantly greater incidence and severity of postoperative shivering than those in a control group. 246Augmented shivering was evident with both glycopyrrolate and hyoscine, suggesting modulation of a mechanism peripheral to the central nervous system. However, a limitation in this study was that the control group was given metoclopramide, a drug possessing parasympathomimetic activity and that is a selective D2-dopamine receptor antagonist. Recent data indicate that atropine slightly increases the thresholds triggering vasoconstriction and shivering, 149which is consistent with the premedication study.
Sites of Action.
There are numerous potential anatomic substrates for the antishivering effect of physostigmine, situated at both supraspinal and spinal levels. In addition to the major cholinergic nuclei and pathways, cholinergic interneurons are found throughout the central nervous system. 247Furthermore, functional interactions between the adrenergic and muscarinic systems are well established. 245,248Most prominently, there are serotoninergic afferents from the raphe nuclei that project to cholinergic brain stem nuclei. 249Through dual projections, cholinergic and aminergic brainstem neurons can concurrently modulate the activity of neurons in the thalamus and basal forebrain during cortical arousal. 250However, the role of these anatomic structures in the thermoregulatory modulation by physostigmine remains hypothetical.
A large number of peptides are found in the brain, especially within the hypothalamus, and there is considerable evidence that they participate in central thermoregulatory control. 251They can be divided into three categories, according to the changes in firing rate of thermosensitive neurons induced by local application of these substances in the preoptic-anterior hypothalamus and the concomitant changes in body temperature.
Local application of thyrotropin-releasing hormone decreases activity of preoptic-anterior warm-sensitive neurons and excites cold-sensitive neurons, thereby producing cold-defense responses and hyperthermia. 252,253In contrast, hypothermia-producing substances (angiotensin II 254and morphine 255) excite and inhibit the activity of preoptic-anterior warm-sensitive and cold-sensitive neurons, respectively. Poikilothermia-producing peptides such as bombesin and neurotensin 256decrease the firing rate in 50–70% of the preoptic-anterior hypothalamic neurons, regardless of their thermosensitivity, with inhibition of both heat-defense and cold-defense responses. Arginine vasopressin, adrenocorticotropic hormone, and α-melanocyte stimulating hormone are thought to act as endogenous antipyretics during fever. 257,258
Opioid peptides induce changes in body temperature that depend on the species, dose, ambient temperature, and degree of restraint during testing. 259Met-enkephalin and β-endorphin induce hyperthermia when given intracerebroventricularly in a low dose, the precise mechanism being unclear. 260At higher doses, enkephalin and β-endorphin cause hypothermia, probably because of a reduction in metabolic heat production. 261,262Microinjected into the preoptic-anterior hypothalamus or the periaqueductal gray, β-endorphin evokes hyperthermia, 263,264as does the injection of enkephalin into the preoptic-anterior hypothalamus. 265Infusion of β-endorphin into the lateral cerebral ventricle of the rat, however, causes hypothermia. 266
Pure μ-receptor agonists, including morphine (2.5 mg), fentanyl (25 μg), and alfentanil (250 μg), may be significantly better treatments for postanesthetic shivering than placebo. 267–269Alfentanil is probably effective because increasing plasma concentrations linearly reduce the shivering threshold. 70,270Epidurally administered sufentanil in patients produces a dose-dependent decrease in shivering response and body temperature. 271Epidural fentanyl also reduced the shivering threshold when added to lidocaine for epidural anesthesia. 272
Meperidine 273is not only an effective treatment for shivering, 274–278but the drug is clearly more effective than equianalgesic concentrations of pure μ-receptor agonists. 70,276Meperidine decreases the shivering threshold almost twice as much as the vasoconstriction threshold (fig. 6). 279This is in distinct contrast to other analgesic and sedative drugs, including propofol, 67dexmedetomidine, 152and midazolam 280(table 1), and to general anesthetics. 68,69The gain and maximum intensity of shivering remain unchanged during both alfentanil and meperidine administration. 70,279These results thus demonstrate that the special antishivering effect of meperidine is primarily mediated by a disproportionate reduction in the shivering threshold.
The antishivering action of meperidine may be partially mediated by κ-opioid receptors. 145Consistent with this theory, nalbuphine and butorphanol, two other antishivering drugs, 140,149,153,281are known to have κ-opioid receptor activity. 282–284The difficulty with this theory is that recent data indicate that nalbuphine, a mixed μ-antagonist and κ-agonist, comparably reduces the vasoconstriction and shivering threshold in volunteers. 149
Sites of Action.
Possible substrates for the effects of opioids on body temperature and thermoregulatory responses include their actions on preoptic-anterior hypothalamus neurons, 285dorsal raphe nucleus neurons, 286raphe magnus neurons, 287locus coeruleus, 288and the spinal cord. 272Generally, opioids exert a variety of stimulatory effects on signal transduction, 289including stimulation of cyclic adenosine monophosphate formation. Cyclic adenosine monophosphate increases thermosensitivity in warm-sensitive and moderate-slope temperature-insensitive neurons. 290,291
A significant increase in temperature sensitivity was observed in warm-sensitive preoptic-anterior hypothalamic neurons during administration of the κ-opioid receptor opioid agonist dynorphin A1–17. 285Selective κ-opioid receptor agonists attenuate the response of locus coeruleus neurons to excitatory inputs; in contrast, the μ-opioid receptor agonist morphine directly inhibits or excites basal locus coeruleus discharge. 292The extent to which κ-opioid receptors in the hypothalamus, spinal cord, 293and locus coeruleus contribute to the thermoregulatory effects of meperidine remains unclear. However, the modest thermoregulatory effects of nalbuphine 149might suggest that mechanisms other than κ-opioid receptor agonism predominate.
The positive ions calcium (Ca2+) and sodium (Na+) may play a functionally opposing role in mediation of body temperature. 161In monkeys, 294perfusion of excess Ca2+into the posterior hypothalamus evokes a decrease in body temperature, whereas perfusion with Na+ions increases body temperature. The magnitude of this response depends on the ratio of the cations’ concentration 295and may thus define the set-point for body temperature. The ratio of these cations in the posterior hypothalamus shifts immediately after an intense peripheral thermal challenge. 202During conditions of fever and defervescence, push-pull perfusate studies confirm that the ratio of the cations changes corresponding with the direction of change in body temperature. 296,297
Less experimental data are available on the possible role of magnesium in the regulation of body temperature. Magnesium may be considered a physiologic calcium-channel blocker. 298Magnesium chloride injected into the third ventricle of the sheep increases body temperature, 299whereas intracerebroventricular injection of Ca2+elicits hypothermia in other species. 300During cold exposure, magnesium concentration in rat plasma increases, 301and in heat-acclimated volunteers, plasma magnesium decreases. 302In addition, during treatment with magnesium sulfate, a significant decrease in maternal temperature was observed. 303The possible physiologic role in cold adaptation may thus explain the effectiveness of magnesium in decreasing the threshold of postanesthetic shivering.
N -methyl-d-aspartate Receptor Antagonists
Magnesium sulfate is a physiologically occurring competitive antagonist at NMDA receptors 293and was recently found to stop postanesthetic shivering. 147Several antishivering drugs share the NMDA receptor antagonist properties of magnesium.
Orphenadrine is both antimuscarinic 304and has noncompetitive NMDA receptor antagonist properties. 305Orphenadrine extends the action of perioperative analgesics 306and thus has been proposed as an alternative to methylphenidate to control postanesthetic shivering. 148
Ketamine, which is a competitive NMDA receptor antagonist, 307also inhibits postanesthetic shivering 143; however, this effect must be interpreted with caution because of the drug's pharmacologic complexity. In addition to being a competitive NMDA receptor antagonist, ketamine has several other pharmacologic properties, 308including being a κ-opioid agonist, 309blocking amine uptake in the descending inhibitory monoaminergic pain pathways, 310,311having a local anesthetic action, and having an interaction with muscarinic receptors. 309
It is likely that NMDA receptor antagonists modulate thermoregulation at multiple levels. There are neurons in the preoptic-anterior hypothalamus of the rat whose firing rate increases by application of NMDA. 312Furthermore, NMDA receptors modulate noradrenergic and serotoninergic neurons in the locus coeruleus. 313,314In the dorsal raphe nucleus, 5-HT acts as a neuromodulator to enhance the effects of NMDA receptors. 286Finally, the NMDA receptors at the dorsal horn of the spinal cord modulate ascending nociceptive transmission. 293The relation between nociceptive transmission and afferent thermoregulatory pathways nonetheless remains largely speculative.
Methylphenidate is effective for prevention and treatment of postanesthetic shivering. 139,315Methylphenidate is an analeptic agent that binds presynaptic sites on dopamine, norepinephrine, and 5-HT transport complexes, which in turn block reuptake of the respective neurotransmitters. 316,317Activation of the raphe system and the concomitant arousal 318may explain the impressive antishivering potency of methylphenidate. However, experimental evidence for the precise anatomic substrate of methylphenidate's antishivering action is lacking.
Doxapram is a low-potency analeptic agent that is best known as a respiratory stimulant. However, it is also an effective treatment for postanesthetic shivering. 142Although the pharmacology of doxapram remains poorly understood, the drug clearly stimulates breathing by a central action in or below the pons as its action is unaffected by brainstem transection in fetal lambs. 319Doxapram speeds up awakening after halothane anesthesia, 320as does physostigmine. 321In dogs recovering from halothane anesthesia, this clinical observation was confirmed by electroencephalographic evidence of arousal after administration of each drug. 322
Does Meperidine Have a Unique Antishivering Mechanism?
Finally, we return to the intriguing question: which pharmacologic properties of meperidine mediate its antishivering action? Meperidine is the only member of the opioid family that has clinically important local anesthetic activity in the dose range normally used for analgesia and is unique among currently used opioids in being effective as a sole agent for spinal anesthesia. 323However, local anesthetic action does not appear to mediate the drug's antishivering action in humans since a clinical dose of intravenous lidocaine does not prevent shivering 324or reduce the shivering threshold. 325
Analgesic concentrations of meperidine produce considerable inhibition of 5-HT reuptake. 326Meperidine, in combination with a monoamine oxidase inhibitor, can consequently cause fatal hyperthermia that is presumably caused by the accumulation of brain 5-HT. 327The 50% inhibitory concentration of meperidine for 5-HT reuptake is 490 nm, but more than 100,000 nm for morphine. 328Moreover, meperidine in analgesic doses is among the most potent inhibitors of norepinephrine reuptake in central neurons 326,329,330and isolated nerve endings. 329,331This effect is not inhibited by naloxone and is therefore not opioid receptor mediated.
Meperidine possesses several other nonopioid actions, 332one or more of which may explain this drug's special antishivering action. For example, meperidine has noncompetitive NMDA receptor antagonist activity in the rat spinal cord. 333Possible mechanisms by which NMDA antagonists interfere with shivering were previously discussed. Finally, does meperidine, as was claimed when it was introduced as an antispasmodic in 1939, have anticholinergic effects? 334In the presence of a parasympathomimetic, meperidine is a competitive antagonist of muscarinic receptors in guinea-pig ileum. 335Furthermore, meperidine shows significant muscarinic receptor binding in mice (K I= 1.7 μm; Elmar Friderichs, M.D., written communication, June 28, 1999) However, recent data indicate that atropine slightly increases the threshold triggering shivering. 149
An important recent contribution to the discussion on the mechanism by which meperidine inhibits postanesthetic shivering was made by Takada et al . 336They transfected COS-7 cells with the cDNA for human α2A, α2B, and α2Cadrenoceptors. Results indicate that meperidine can bind to each of the α2-adrenoceptor subtypes and transduces an agonist action at these sites. The α2B-adrenoceptor subtype is the most sensitive and thus appears to be the most important receptor subtype in this regard. The next step is linking these pharmacologic findings to anatomic structures. However, the results are consistent with the possibility that meperidine exerts some antishivering action via α2adrenoceptors in the locus coeruleus.
Meperidine thus possesses special antishivering properties that are not shared by pure μ-receptor opioids. This special antishivering action is mediated by a shivering threshold that decreases twice as much as the vasoconstriction threshold 279throughout the range of tested doses (table 1). κ-Opioid receptor agonists have antishivering effects, but nalbuphine comparably inhibits vasoconstriction and shivering 149—suggesting that κ-opioid activity does not explain the special antishivering action of meperidine. Although meperidine has an anticholinergic action, this also does not appear to be the explanation for its singular antishivering efficacy. However, the explanation may involve its biogenic monoamine reuptake inhibition, NMDA receptor antagonism, or stimulation of α2adrenoceptors. The special antishivering action of meperidine may simply result from the drugs's lack of specificity and a fortunate accumulation of pharmacologic actions modulating thermoregulatory shivering.
Because hypothermia is associated with shivering and so many other complications, surgical patients should be kept normothermic 337unless hypothermia is specifically indicated for putative protection against cerebral ischemia 338–341or spinal cord injury. 342Given the discomfort and metabolic stress associated with shivering, treatment is appropriate in most cases. Any effective treatment for shivering will, by definition, reduce metabolic heat production and must be accompanied by an effective active heating system or a high ambient temperature. 343
An inventory of the known antishivering drugs reveals striking similarities in their pharmacologic properties (table 2). However, conclusions should be cautiously extracted from this overview, because several drugs possess these mechanisms without any (known) thermoregulatory effect. Moreover, these common features are interrelated. Almost all antishivering drugs, for example, produce a transient vasopressor response. 243,308,318,319,344–347On a theoretical basis, one cannot exclude that the presence of norepinephrine in the blood, resulting from a spillover from neuronal synapses, further increases the inhibition of cold defenses immediately after intravenous injection of the drug. Circulating catecholamines modulate the static and dynamic activities of skin cold receptors. 348,349
In a classic article, Satinoff 23postulated that thermoregulatory reflexes evolved out of systems that were originally used for other purposes, called “evolutionary coadaptation.” He argued that it would be unnecessarily burdensome to require the evolutionary process to create new systems to solve a problem already solved by an existing system. For example, the peripheral vasomotor system first served as a supplemental respiratory organ in amphibians. It then became a heat collector and disperser in reptiles, and finally an essential thermoregulatory control mechanism in mammals. Similarly, the muscular organization in reptiles and the consequent changes in posture provided the basis for an internal heat production in mammals.
In place of the commonly held view of a single thermoregulatory integrator (i.e. , the preoptic area of the hypothalamus) with multiple inputs and outputs, modern concepts include integrators for each thermoregulatory response. 22Furthermore, these integrators are distributed among numerous levels within the nervous system, with each being facilitated or inhibited by levels above and below. 23
All antishivering drugs except ketanserin have weak or moderate analgesic properties in humans. The descending pain-control network acts pharmacologically through biogenic monoamines, and there is thought to be considerable interaction between antinociceptive and thermoregulatory systems. 55,215,350Central aminergic systems exert a general modulatory influence on neurons involved in different functional and neuroendocrine systems. 39,351
It thus seems reasonable to assume that thermoregulation is tightly linked to other homeostatic systems, including the control of pain. Pain and temperature signals are transmitted along similar fiber systems that synapse in dorsal horn regions. As mentioned previously, electrical stimulation of the rostral ventromedial medulla not only causes an increase in the analgesia to noxious stimuli, but also a decrease in the thermoregulatory response to peripheral warming and cooling. 215,352,353One of the important functions of the rostral ventromedial medulla is to modulate the amount of pain and temperature input ascending from the spinal cord by gating the transmission of neuronal signals at the level of the dorsal horns. 137This interesting expansion of the existing pain and thermoregulatory control models deserves further experimental investigation.
In summary, it is difficult to link pharmacologic properties to anatomic substrates and, specifically, to the control of thermoregulatory shivering. Even a partial understanding of the mechanisms involved in the shivering response reveals an extraordinary complexity, presumably the result of evolutionary coadaptation. No single structure or pathway is responsible for mediation of the thermoregulatory shivering response. In contrast, several mechanisms are able to modulate various thermoregulatory responses.
The authors thank Elmar Friderichs, M.D. (Department of Pharmacology, Grünenthal GmbH, Aachen, Germany), and colleagues for discussing pharmacologic topics in this manuscript.