Barbiturates

Neil Harrison, Wallace B. Mendelson and Harriet de Wit

From their introduction into clinical practice at the beginning of the 20th century until recent years, the barbiturates have enjoyed a central place in the pharmacopoeia of CNS drugs. Until the benzodiazepines were introduced in the 1960s, barbiturates were widely used clinically for a range of indications, including the treatment of anxiety, insomnia, seizure disorders and as muscle relaxants and anesthetic agents. Benzodiazepines and the newer non-benzodiazepine hypnotics are now preferred over barbiturates for most of these clinical uses because they have a wider therapeutic index, tolerance develops more slowly, and their liability for abuse is lower than that of the barbiturates. Nevertheless, barbiturates remain an important class of drugs from a scientific point of view, because they have played a central role in the characterization of the GABAA receptor complex. This chapter will focus in particular on current knowledge and recent developments in our understanding of the receptor actions of barbiturates. We will also review the clinical pharmacology of this class of drugs, with particular attention to similarities and differences between barbiturates and benzodiazepines and their effects on sleep.

No drug has a single action. Perhaps no class of drugs better illustrates this most important of axioms in pharmacology than the barbiturates. The barbiturates have sedative-hypnotic, anticonvulsant, anesthetic and respiratory depressant effects that are mediated by their actions at various target sites in the body. Despite the different mechanisms, however, there is considerable overlap between the therapeutic/toxic dose-response curves with these drugs. For example, the anticonvulsant effects of phenobarbital are associated with significant sedation (53), and the anesthetic effects of pentobarbital with respiratory depression.

Researchers have attempted to identify the actions of barbiturates in the CNS since the discovery of this class of drugs. Early attempts to identify the basis of their CNS actions related to the hydrophobic nature of the drugs and their possible interactions with membrane phospholipids, interference with cellular energy metabolism and intracellular free ionized calcium levels (reviewed in ref. 28). All of these concepts eventually fell victim to a second important pharmacological principle: drug mechanisms in vitro are only important if they occur over a concentration range relevant to the therapeutic/toxic effects under consideration. For example , the fluidizing actions of barbiturates on membrane phospholipids are rather small until supratherapeutic concentrations are reached. More recently, neurophysiologists have pursued the idea that the site of general anesthesia/CNS depression could be a protein. This idea originated with researchers studying neurotransmitter receptors and ion channels involved in the control of CNS excitability (reviewed in ref. 18). Several neuronal targets are affected by pharmacologically relevant concentrations of barbiturates, including ligand-gated and voltage-gated ion channels. The role of these different potential targets in synaptic transmission is summarized in Figure 1). Each potential target will be reviewed briefly here; at the end we will report in detail on recent progress in elucidating the molecular mechanisms of barbiturate interactions with the GABAA receptor.

Postsynaptic inhibition in the brain, is usually mediated by GABA (38). Action potentials generated in the inhibitory interneuron trigger release of GABA, which binds to postsynaptic receptors and causes the opening of chloride ion channels in the postsynaptic membrane. A transient change in postsynaptic membrane potential (inhibitory postsynaptic potential; IPSP) and a fall in input resistance results, stabilizing the postsynaptic membrane potential below threshold. Presynaptic inhibition also occurs in the CNS, in which case the output of excitatory neurotransmitter is reduced by the action of an inhibitory transmitter acting at a presynaptic receptor on the excitatory nerve terminal. Eccles et al. (16) noted that GABA-mediated presynaptic inhibition of cat motoneurons was prolonged by barbiturates. Later it was shown that hippocampal postsynaptic inhibition is also prolonged by pentobarbital (45,60). Such increases in pre- and postsynaptic inhibition are invariably associated with decreased neuronal firing and network activity. These findings led naturally to a focus on interactions of barbiturates with the GABAA receptor. This 'classical' GABA receptor is blocked competitively by bicuculline and non-competitively by picrotoxin. It also possesses a number of important allosteric modulatory sites at which various drugs act. There is a distinct site for noncompetitive GABA antagonists, such as picrotoxin, and a relatively well-characterized binding site for anxiolytic and hypnotic benzodiazepines (BZ). In addition, the GABAA receptor is modulated at unknown sites by barbiturates. The imidazodiazepine flumazenil antagonizes all of the actions of the BZs, but fails to influence those of the barbiturates (reviewed in ref. 49), demonstrating that BZs and barbiturates act at distinct allosteric sites on the GABAA receptor macromolecule.

The effects of barbiturates on the GABAA receptor have been extensively studied using electrophysiological techniques in vivo (11) and in isolated or tissue-cultured neurons, in which barbiturates enhance the actions of submaximal GABA concentrations (1,6), by prolonging the openings of individual GABA-operated chloride channels (35,65). The potency of various barbiturates as modulators of the GABAA receptor correlated well with their hydrophobicity and with their potency as hypnotics and anesthetics (27). Radioligand binding studies also revealed allosteric interactions of the barbiturates at the GABAA receptor (5,46), with barbiturates causing an increase in the binding of labeled GABA and benzodiazepines to the receptor.

Barbiturate effects on GABAA receptor function have been studied at the single-channel level in cultured mouse spinal neurons (35). Openings of the GABA-activated chloride channels occur in 'bursts', interrupted by frequent brief closures. The channel kinetics are best described by a complex model involving three open states, having open time constants of 1, 4 and 11 msec. As the concentration of GABA is increased, the channels open more frequently and exhibit longer open times, entering the longest-lived open states more frequently. In the presence of 50 mM pentobarbital, the main conductance state, opening frequency, and individual open time constants are not altered, but the frequency of occurrence of the longest open state increases, as does the occurrence of long bursts generated by that long open state (35). Barbiturates therefore promote entry of GABA-activated channels into a long-lived open state (35), whereas BZs increase only the frequency of channel opening into the initial open state (65,69). These mechanistic studies reveal interesting details of the changes in channel gating caused by barbiturates but as yet have yielded no insights into the molecular sites of action. An additional interesting effect of barbiturates is direct gating of the channels, i.e., the barbiturates may open the channel even in the absence of GABA (1,2). This usually occurs at significantly higher concentrations than those which potentiate the actions of GABA; these concentrations also are generally higher than those required for clinically effective anesthesia.

L-Glutamate is now known to be the universal fast excitatory neurotransmitter of the mammalian central nervous system. There are multiple receptor subtypes for L-glutamate, and these can be broadly subdivided pharmacologically into AMPA- (or quisqualate), N-methyl D-aspartate (NMDA) and kainate receptors, according to their preferred selective agonists among amino acid analogs of L-glutamate. As the major central excitatory transmitter, glutamatergic synapses are an obvious target for the actions of CNS depressant drugs such as the barbiturates. Richards first noted the depression of EPSPs by pentobarbital in slices of olfactory cortex (55) and subsequently demonstrated that pentobarbital also depressed the excitation of neurons in the prepyriform cortex by L-glutamate (56). Similar effects were noted in tissue cultured neurons (6). In sodium flux measurements, barbiturates depressed flux induced by quisqualate but not by NMDA (67). A study in cortical slices showed that pentobarbital depressed responses to quisqualate and kainate but not those elicited by NMDA (26), indicating a receptor-subtype-selective action of the barbiturate. Studies in cultured neurons confirmed these findings. The block of the channels associated with the AMPA receptor appears to be both voltage- and use-dependent (37), indicating penetration of the barbiturate deep into the ion conducting pore. Cloning of the glutamate receptor subunit cDNAs has enabled investigators to explore this problem at the molecular level, where it was noted that the potency of pentobarbital as a glutamate antagonist is critically dependent upon the result of editing of the crucial Gln/Arg site in the pore-forming loop of the Glu-R subunits. A specific RNA editing enzyme can interconvert a glutamine or arginine codon at a position which is also crucial to the ion selectivity and permeation of the glutamate-operated ion channel.

Voltage-dependent sodium channels allow sodium to enter the cell in response to membrane depolarization. These channels are thus crucial in the generation and conduction of the action potential. Although undoubtedly the primary site of action of local anesthetics, the voltage-gated sodium channel has not been thought to be of primary importance in the CNS depressant actions of the barbiturates. Axonal conduction in peripheral nerves is quite unaffected by anesthetic concentrations of the barbiturates, for example. However, various anticonvulsant drugs, including phenobarbitone, have been shown to limit extremely high-frequency firing of neurons (36), and these effects are likely to involve voltage-gated Na+ channels. Some interesting alterations in the voltage-dependent gating of sodium channels have been reported (19). Although these effects generally occur at concentrations beyond the therapeutic range, the concentration-dependence of barbiturates may vary with sodium channel subtypes. Novel brain sodium channels have been revealed by cloning techniques, and these may have higher sensitivity to the barbiturates than those channels studied to date.

Voltage-dependent potassium channels allow potassium to exit the cell, often in response to membrane depolarization. These channels are therefore usually involved in repolarization following excitation or in regulating and controlling the resting membrane potential. Researchers have identified and cloned a seemingly infinite number of potassium channels. Barbiturates inhibit some voltage-gated potassium channels, notably the large conductance calcium-activated K+ channels (43,48), and some inwardly rectifying K+ channels (20). However, the concentrations of barbiturates responsible for inhibiting K+ channel function tend to be relatively high, and—in general—such actions would lead to increases in neuronal excitability, thus opposing the general trend toward CNS depression by these drugs. It is possible that the paradoxical effects of certain barbiturates (e.g., methohexital [Brevital]) can precipitate epileptiform discharges in the EEG of sedated epileptic patients) might result from interference with K+ channel function. In fact, some barbiturates are potent convulsants (14), and yet these compounds are known to share the potentiating actions of pentobarbital at GABAA receptors (27), rather than to act as GABA antagonists, as do certain other convulsants. Clearly, the convulsant barbiturates must possess potent actions at CNS targets which override their effects on GABAA receptors. K+ channel mechanisms might bear closer inspection in this context, since the mechanisms of the convulsant barbiturates remain quite unexplained.

Voltage-dependent calcium channels allow calcium ions to pass into cells in response to membrane depolarization. Many calcium channel subtypes have been identified, with L-type channels predominant in heart and skeletal muscle, but a complex array of N, T, P, Q and R types identified in neurons (76). Barbiturates inhibit the flow of calcium through several types of voltage-gated calcium channels, and blockade of L-type Ca2+ channels is likely to explain in part their sometimes profound cardiac depressant actions (22). Neuronal N-type calcium channels are also barbiturate-sensitive (17), and inhibition of these channels may be relevant to the presynaptic effects of barbiturates to decrease the output of excitatory transmitter at certain synapses (72). An explosion of activity in cloning calcium channel subtypes has taken place since these studies on neuronal calcium currents, and so a re-examination of the effects of barbiturates on various cloned channel subtypes is to be anticipated.

The GABAA receptor is a member of a 'superfamily' of related ligand-gated ion channels, along with the nicotinic acetylcholine receptors. These receptors are pentameric subunit complexes arranged around a central permeation pore (Figure 2). Each subunit has four putative transmembrane (TM) domains; the second TM domain (TM2) is thought to form the lining of the pore (75), which has a diameter of ~5.4 Å (7).

The first GABAA receptor subunit cDNAs were cloned in 1987 (61), and many more have been cloned since (review in ref. 47). In general, combinations containing abg subunits behave more like native GABAA receptors (4,70). Indeed, the stoichiometry of native GABAA receptors is now thought to involve two a subunits together with two b subunits and one g subunit (9). GABAA receptors consisting of only a and b subunits may be formed in expression systems, however, and these show barbiturate sensitivity (34). The g subunit is essential to confer diazepam sensitivity to the receptor (51), but it is not required for modulation of GABAA receptors by the barbiturates (23,34,50). d and e subunits have also been reported (12).

The rich variety of a subunit isoforms provides considerable pharmacological diversity. Six different a subunit isoforms have been reported, along with three isoforms of the b subunit. The a subunit isoform influences agonist potency (34) and BZ pharmacology (24,51). Alpha-6 or a4-subunit - containiry receptors are completely insensitive to BZ agonists (24,73). In contrast to the importance of the a subunit, variation of the b subunit isoform has little influence on the barbiturate pharmacology of recombinant GABAA receptors (25). Variation of a subunit isoform alters the efficacy but not the potency of barbiturate modulators (33,68), indicating that the variable regions of the a subunit are not likely to be the primary binding site for the barbiturates. The Tyr-Gly-Tyr sequence in the b subunit is undoubtedly important for the binding of GABA and other direct receptor agonists (3), as is Phe 64 in the a1 subunit (62). The g subunit appears essential for stabilizing BZ binding to the a subunit (64). A single histidine residue (replaced by Arg in a6) is of vital importance for BZ agonist binding (73). However, neither the Tyr-Gly-Tyr sequence in b, the Phe 64 in a, nor the His 100 in a is of vital importance for barbiturate function (3). Some reports suggest that the b subunit alone may form a site for barbiturates (59). A novel approach to this question employed chimeras between GABAA and glycine receptors (which are virtually insensitive to barbiturates). These studies demonstrated no requirement for either N- or C-termini, nor for the large cytoplasmic loop or TM4 of the GABAA receptor subunits in the actions of the barbiturates (32). abe combinations are reported to be insensitive to potentiation by barbiturates (12). Further studies of this type should uncover the barbiturate sites of action on this receptor in the near future; it is clear that these are different from the critical amino acid residues for the actions of the anesthetics isoflurane and propefer at the GABA (33a) receptor.

The study of the molecular mechanisms of the action of barbiturates has progressed from the original work on targets identified in the 1970s and 1980s to the application of molecular biology in the 1990s. In the next century, we are likely to see the molecular approaches come to fruition, yielding descriptions of barbiturate sites of action as detailed as those now available for the local anesthetic block of voltage-gated sodium channels, for example (52). In addition, the advent of transgenic, knock-out, 'knock-in' and tissue-specific knock-out technologies will enable the alteration or ablation of these targets in live animals, which can then be tested for their pharmacological responses to barbiturates. Indeed, knockouts of GABAA receptor g2 and b3 subunit genes have already been performed (23,30). These experiments hold great promise to resolve the most important pharmacological questions regarding which molecular species are important for the anticonvulsant, sedative, anesthetic, and toxic actions of the barbiturates.

The barbiturates were the most widely used sedative/hypnotics from the early part of this century until the early 1970s, when flurazepam (the first benzodiazepine specifically recommended for sleep) entered the US market. The agents most commonly used as hypnotics are the short- to intermediate-acting compounds such as amobarbital, pentobarbital and secobarbital, which have half-lives of 10-15 hours to 40-50 hours. Efficacy studies of barbiturates on sleep are reviewed by Mendelson (40,41). Barbiturates produce the classical (indeed, they define the classical) effects of hypnotics, which include shortened sleep latency, increases in total sleep, and often a decrease in waking time during the night. Whereas the barbiturates have inconsistent effects on slow-wave sleep, they consistently and potently depress rapid eye movement (REM) sleep. Benzodiazepines, in contrast, are potent suppressors of slow-wave sleep but have relatively mild REM-suppressing effects. The lesser effect of benzodiazepines on REM suppression was at one time viewed as advantageous. However, later researchers have questioned the degree to which REM suppression is harmful, or whether it is harmful at all and indeed REM deprivation studies have indicated that in some situations it may even be therapeutic (i.e., as a treatment for depression). Moreover, since the functions of the various sleep stages remain uncertain, at this time it is not clear whether there are advantages to relative reductions in one specific stage, relative to another. Abrupt cessation of recommended doses of hypnotic barbiturates, as with shorter-acting hypnotics, leads to transient sleep disturbance. This is often accompanied by temporary increases in the amount of REM sleep. Once again, it is not clear whether the elevated duration of REM sleep specifically translates into the subjective experiences of distressed sleep.

During sleep, barbiturates reduce neurogenic respiratory drive; doses approximately three times those used therapeutically virtually eliminate neurogenic drive and greatly reduce hypoxic drive (29). Protective respiratory reflexes such as coughing are only mildly affected until very high doses are administered. For these reasons, patients with pre-existing respiratory compromise (e.g., sleep apnea) may be at increased risk when given hypnotic barbiturates. Cardiovascular effects of oral hypnotic barbiturates are minimal and usually confined to mild reductions in blood pressure during sleep (29). Orally administered hypnotic doses do not affect the rate of gastric emptying. Hepatic effects include induction of microsomal enzymes responsible for the metabolism of many other drugs as well as endogenous compounds such as steroids, cholesterol and some vitamins. Mitochondrial and cytoplasmic enzymes may also be affected. Oral hypnotic doses have virtually no analgesic properties, suggesting that they are not helpful (at least when given alone) in patients for whom sleep disturbance is secondary to pain.

The acute effects of barbiturates and benzodiazepines on mood and subjective state have been well documented both in healthy volunteers and in individuals with histories of drug abuse (13). Acute doses of barbiturates decrease anxiety and increase feelings of fatigue, dizziness, lightheadedness and lethargy. In individuals with a history of drug abuse, they also increase self-reported feelings of being "high." Some of these subjective experiences may be therapeutically desirable, such as the anxiolytic effect when the drug is used as a preoperative medication, or they may be undesirable, as in the case of the sedative effect when the drug is used as an anticonvulsant. The subjective states produced by sedative drugs are also closely associated with their abuse or non-medical use. Interestingly, sedative effects of barbiturates and benzodiazepines are considered pleasurable by individuals who have a history of drug abuse, while the same or similar effects are considered unpleasant by most individuals who lack an extensive history of drug use (13,74). The subjective effects of acute drug administration, especially ratings of 'liking' of the effects, are thought to be a good indicator of a drug's abuse liability (31). Laboratory studies have also been conducted to assess behavioral preference for barbiturates over a placebo, using double-blind choice tests. In these studies, healthy volunteers typically prefer an inactive placebo over a barbiturate (10), whereas individuals with a history of drug abuse typically prefer barbiturates over both placebo and comparable doses of benzodiazepines (21).

Barbiturates, like benzodiazepines, impair performance in a dose-dependent manner on a wide range of psychomotor tasks. The psychomotor tasks typically used in laboratory studies represent components of the skills that are required for complex, coordinated activities such as driving a car. Barbiturates impair performance on standard tests of eye-hand coordination such as circular lights and the digit symbol substitution tests, and they decrease the number of items recalled in memory recognition tests (57). In most respects, the psychomotor effects of barbiturates resemble those of the benzodiazepines. Interestingly, however, benzodiazepines have more pronounced effects on memory than barbiturates, at doses that produced comparable subjective feelings of sedation (57). In addition, subjects under the influence of benzodiazepines underestimate their level of impairment on psychomotor tasks, compared with subjects under the influence of barbiturates. This relative lack of awareness of the motor impairment is likely to increase the risks of performing tasks requiring concentration or dexterity (e.g., driving) under the influence of benzodiazepines.

Barbiturates still have certain therapeutic and diagnostic uses. Phenobarbital continues to be used in the treatment of seizure disorders, and the shorter-acting barbiturates are a useful adjunct, or occasionally even the primary agent, for anesthesia. In addition, barbiturates are used as a diagnostic procedure prior to neurosurgery. This may involve administration of methohexital to localize epileptic foci, or it may involve intracarotid administration of sodium amobarbital (71). The sodium amytal procedure produces a transient unilateral suppression of hemispheric function, and it is the definitive diagnostic procedure to determine hemispheric dominance for speech and language. It is used to localize critical language and memory functions that should be spared in neurosurgical procedures. It has been used effectively for almost 50 years, and continues to be used by many centers today (54,63).

The barbiturates are highly toxic in acute overdose. Ten times the therapeutic dose may lead to fatal respiratory or cardiovascular depression. Patients who survive may develop renal failure secondary to the transient anoxia experienced during the toxic state. This severe toxicity profile emphasizes the importance of considering the possibility that patients who present with sleep dysfunction may be clinically depressed. If depression is not recognized, administration of a hypnotic (particularly a barbiturate) may thus provide the patient with a means of committing suicide, as well as denying him/her the appropriate medication (i.e., an antidepressant). In a study in which toxicologies were performed in over 90% of 204 consecutive suicides seen by the San Diego County Coroner in the early 1980s, anxiolytics and hypnotics were found in 10.7% and 12.3% of cases, respectively (41). Barbiturates and benzodiazepines were found in roughly equal proportions (7.5% and 9.6%, respectively), although at that time there were only one-sixth as many barbiturate prescriptions filled at drugstores nationally. An examination of 158 intentional overdose patients who were seen during 1994-1995 at the Cleveland Clinic found that prescription and over-the-counter sedative/hypnotics accounted for 16.4% and 6.3% of cases respectively (Mendelson, unpublished observation). The most common sedative/hypnotics taken were diazepam and alprazolam, and in this series there were no cases of intentional barbiturate overdose, perhaps reflecting the decrease in their administration nationally.

The most common adverse reaction after recommended doses is drowsiness on waking, which occurs only rarely the following morning after bedtime administration. Paradoxical excitement may be seen in the elderly and debilitated, as well as in patients who are in pain. Perhaps of greatest concern are drug interactions, in particular the potentiation of sedation by other agents, especially ethanol and benzodiazepines. Acutely, barbiturates can competitively inhibit the metabolism of other drugs or endogenous compounds, and, conversely, during chronic administration they may increase the rate of hepatic clearance of other compounds which are handled by the same drug-metabolizing systems. A study of reported adverse reactions to sedative/hypnotics during a three-year period in a 1000- bed teaching hospital found no reported reactions to pentobarbital (546 doses dispensed) and only one case (a hypersensitivity skin rash) from 21,531 doses of phenobarbital (42).

Abuse of sedative-hypnotics, including the barbiturates, is low relative to other classes of abused drugs (44). Moreover, abuse of barbiturates has declined substantially over the past 20 years, perhaps because of decreased availability related to declines in prescription use. One estimate of the prevalence of abuse is the annual survey of secondary school students in the Monitoring the Future Study, sponsored by the National Institute on Drug Abuse. This survey shows that the percentage of 12th grade students who reported using barbiturates in the previous 30 days declined from 4.7% in 1975 to 1.7% in 1994. This 64% decline in barbiturate use can be compared to the 15% decline in cigarette use over the same time period and a 29% decline in the use of all illicit drugs. The absolute level of barbiturate use is low, relative to other drugs. For example, in 1994, 1.7% of 12th graders reported use of barbiturates in the last 30 days, whereas 19% reported using marijuana, 31% used tobacco cigarettes, and just over 50% used alcohol. Although estimates of drug use among high school students are usually related to estimates of drug use in the general population, use among high school students may be higher for certain drugs. Another NIDA survey, the National Household Survey, estimates non-medical drug use in the non-institutionalized adult US population. By this measure, in 1992, only 0.4% of respondents reported using sedative drugs in the last 30 days. This low prevalence of sedative use may be compared to 4.8% who reported use of marijuana, 27% who used tobacco cigarettes, and 51% who used alcohol in the last 30 days (66). Thus, data from both high school students and adults indicate that current rates of abuse of barbiturates are low, especially when compared with other contemporary drugs of abuse.

Barbiturates have historically played an important role in the treatment of a variety of disorders, including anxiety disorders, sleep disorders, seizure disorders and muscle spasm, and they have proven useful in anesthesia and in localizing brain dysfunction prior to neurosurgery. Although their use in clinical practice has mostly been replaced by the safer benzodiazepines, they are still used in certain clinical situations. Recently, major advances have been made in our understanding of the receptor mechanisms and molecular pharmacology underlying the actions of barbiturates in the brain. Barbiturates have their primary actions on the GABAA receptor, but they also interact with glutamate receptors and voltage-gated ion channels. Because of the impressive recent advances in molecular biology techniques, our understanding of the GABA receptor and the actions of barbiturates on the central nervous system is likely to be significantly improved in the next decade.

published 2000