Phenobarbital in Status epilepticus – Rediscovery of an effective drug

.


Introduction
In 1864 the German chemist and Nobel Prize winner Adolf von Baeyer (1835-1917) synthesized barbituric acid (malonylurea). Although barbituric acid had no specific effects on the central nervous system (CNS) it led to the successful development of a series of CNS-active barbiturates and structurally related antiseizure medicines (ASMs) [1]. Another German Chemist and later Nobel Laureate, Emil Fischer (1852-1919), and his collaborator Joseph von Mering (1849-1908), discovered the more lipophilic barbituric acid derivative barbital (5,5-diethylbarbituric acid) in 1903, and phenobarbital (5-ethyl-5-phenylbarbituric acid, Fig. 1) in 1911 [2,3]. Both barbital (Veronal Ò ) and phenobarbital (Luminal Ò ) were introduced into the market by the Bayer company (Farbenfabriken vormals Friedredrich Bayer et Co.) as hypnotic and sedative drugs in 1904 (the history of barbiturates is reviewed in:1, 2, 3). It was the German Neuro-Psychiatrist Alfred Hauptmann (1881Hauptmann ( -1948, who discovered the antiepileptic effects of phenobarbital in 1912 [4]. Hauptmann worked at that time at the Neuropsychiatric University Hospital in Freiburg im Breisgau, Germany (Fig. 2a). In his short report, which was published in the ''Münchner Medizinische Wochenschrift", he first praised the excellent utility of phenobarbital as a sedative treatment for states of anxiety, agitation (''Erregungszustände"), and insomnia. He also found the subcutaneous application of its sodium salt very useful. Then, he described the experiences of his own systematic trial in bromide-resistant epilepsy patients. Hauptmann selected only those, which had moderate to severe forms of epilepsy and who had been under observation to have reliable seizure diaries. He used doses of 150 to 300 mg per day and observed a twofold effect of phenobarbital on epilepsy: firstly, it led to a remarkable reduction of seizure frequency, and secondly, the seizure severity decreased. Patients had not only fewer bilateral tonic-clonic seizures (BTCS) but also ''milder seizure types". He also reported a reduced frequency of ''-Dämmerzustände" (dreamy states, or Dämmerattacken), which can be interpreted as non-convulsive Status epilepticus (NCSE), after (bilateral or generalized) tonic-clonic seizures [5,6]. Patients were much less sedated, compared to bromides, and had fewer side effects. A seizure diary of a successfully treated patient was given as an illustrative example (Fig. 2b). Unfortunately, as very often at this time, the total number of patients or any other details were not reported [4]. Nevertheless, this was the birth of phenobarbital as ASM, which soon replaced bromides in most patients in continental Europe and in North America and heralded a new area of non-sedating ASMs, based on chemical and structural modifications of phenobarbital, which followed in the next decades [1].

Pharmacology of phenobarbital
Phenobarbital (Fig. 1) is a weak acid (pK a = 7.3) that is poorly water-soluble (1 mg/ml). It is a white crystalline material with a melting point of 176°C and a molecular weight of 232.23 [7]. The sodium salt of phenobarbital has a better water solubility and has therefore been used for subcutaneous preparations [8]. The phenobarbital preparation for intravenous and intramuscular injection is not an aqueous solution but contains 20% sodium salt of phenobarbital in a mixture of 90% propylene glycol, 10% ethanol (96%), and water for injection, at a pH of 10 to 11 [9]. At physiologic blood pH (7.3) phenobarbital is 50% ionized and 50% nonionized, but the ratio changes accordingly during metabolic acidosis in SE [7,[10][11][12] facilitating the entry into the brain during status. For instance, Simon et al. found a 2.5 times increased brain concentration of phenobarbital in freely convulsing rats as compared to controls [13]. After intravenous injection phenobarbital distributes in healthy individuals in two phases, which can be characterized by a two-compartment mathematical model [8,14,15]. In the early phase, it follows the highly vascularized organs, such as the liver, kidney, and heart. During the late phase, it distributes to the brain, muscles, and intestine.
The average volume of distribution of phenobarbital ranges between 0.54 and 0.73 L/kg in adults [7,16]. Newborns and young infants have a larger volume of distribution ranging from 0.8 to 1 L/ kg [17][18][19]. Due to its low lipid solubility phenobarbital does not accumulate in fat tissue [7,[10][11][12]. These properties may make look like a disadvantage for emergency treatment, but preclinical studies and human trials suggest, that brain uptake in status epilepticus is much faster than in the healthy brain. Phenobarbital is approximately 55% bound to albumin [20]. Protein binding is reduced in neonates; data in the elderly, in whom an age-related decline in albumin often occurs [21], are missing [7].
At steady state, the CSF concentration is similar in adults and infants ranging between 43% and 60% of plasma concentrations [22,23]. Animal experiments suggest that maximal entry takes 12 to 60 minutes [24,25]. The penetration of phenytoin, phenobarbital, and diazepam into the parenchyma of the cat's brain was assessed after IV infusion [25]. Maximum brain concentrations were reached at 1 minute with diazepam, 3 minutes with phenobarbital, and 6 minutes with phenytoin. Phenobarbital and phenytoin remained bound with similar brain concentrations at 1 minute and 60 minutes [25]. However, brain uptake in SE is faster than in healthy animals: Simon et al. studied the brain penetration of phenobarbital in bicuculline-induced prolonged SE (>3 hours) in paralyzed and ventilated sheep [26]. Much higher brain concentrations were achieved during the first 30 minutes of SE, when systemic hypertension, increased cerebral blood flow (CBF), increased peripheral vascular resistance, a fall in brain pH, and an elevation in brain lactate concentrations are present. The highest concentrations were achieved during the first 5 minutes of SE and remained high for 3 hours. In the late phases of status, the brain uptake of phenobarbital was similar to normal controls [26]. In sum, these findings suggest, that phenobarbital is more rapidly entering the brain during status and is there preferentially concentrated for a long time, due to (a) hyperperfusion in the early stages, (b) systemic hypertension with increased mean arterial pressure, and (c) a decrease on pH due to metabolic acidosis and lactate accumulation.
Phenobarbital is eliminated by hepatic metabolism and renal excretion. Cytochrome P450 CYP2C9 plays a major role in the metabolism, whereas CYP2C19 and CYP2E1 only play minor roles; both have no clinical relevance in the treatment of SE. Elimination half-life in adults ranges from 50 to 150 hours after a single intravenous injection [7,8,14,22]. In the first weeks of life elimination of phenobarbital is much slower resulting in a half-life of 77 to 404 hours, decreasing in older children to about 60 hours [7].

Phenobarbital's mechanism of action
Phenobarbital enhances inhibition by interaction with the GABA A -Receptor [27][28][29][30]. Sub-anesthetic concentrations of barbiturates also can reduce glutamate-induced depolarizations [29] via the AMPA subtypes of glutamate receptors sensitive to kainate or quisqualate [31], At higher concentrations that produce anesthesia, pentobarbital suppresses high-frequency repetitive firing of neurons, apparently as a result of inhibiting the function of voltage-dependent, tetrodotoxin-sensitive Na+ channels [32].
Despite their GABA-ergic properties, barbiturates act quite differently from benzodiazepines: Although barbiturates, like benzodiazepines, also enhance the binding of GABA to GABA A receptors in a chloride-dependent and picrotoxin-sensitive fashion, they promote (rather than displace) the binding of benzodiazepines [33]. Barbiturates potentiate GABA-induced chloride currents by prolonging periods during which bursts of channel opening occur, as opposed to benzodiazepines, which increase the frequency of these bursts [34]. Furthermore, in contrast to benzodiazepines, only a and b (not c) subunits are required for barbiturate action [33].
Barbiturate-induced increases in chloride conductance are not affected by the deletion of the tyrosine and threonine residues in the b-subunit that govern the sensitivity of GABA A receptors to activation by agonists [35]. Another potentially important mechanism of barbiturates in status epilepticus is on calcium channels: phenobarbital and pentobarbital inhibit high-voltage activated calcium currents at therapeutic serum concentrations in guinea pig hippocampal preparations [36] (see Table 1).

Preclinical evidence for the use of phenobarbital in Status epilepticus
The differential selectivity of several barbiturates was studied in a mouse model. Raines et al. evaluated the ''anticonvulsant" and ''neurotoxic" effects of thiopental, pentobarbital, butabarbital, phenobarbital, diphenylbarbiturate, and barbital using the maximal electroshock test and the pentylenetetrazol-induced SE model (90 mg/kg and 200 mg/kg) for efficacy and the Rotorod Test for tolerability [37]. Phenobarbital and diphenylbarbiturate exhibited the most favorable protective indices defined as the ratio between the ED 50 and the ND 50 (neurotoxic dose): 2.71 to 3.41 for phenobarbi-tal and 3.85 to 5.0 for diphenylbarbiturate. The treatment of SE with phenobarbital was studied in a rodent model [38]: systemic homocysteine thiolactone was applied to rats with actively epileptogenic cobalt lesions in the left frontal cortex (Cobalt/Homocysteine Model; 40). 130 male Sprague-Dawley rats received intraperitoneal phenobarbital at doses of 0, 1, 4, 16, 32, 64, and 96 mg/kg. The ED 50 values for control of seizures increased from 14.2 mg/kg for control of tonic-clonic seizures to 76.6 mg/kg for control of all motor and electrographic seizures. Effective serum concentrations above 20 lg/ml Serum levels were achieved at 7.5 min after injection with no decrease until 4 h after injection, suggesting a faster entry into the seizing brain and a slower clearance from the brain, than in healthy control animals [38]. The longterm outcome and the development of chronic epilepsy after SE was studied in limbic status epilepticus by ''continuous" hippocampal stimulation in rats receiving phenobarbital, MK-801, and phenytoin at 1, 2, and 4 hours after SE onset [39,40]. Phenobarbital was most effective in suppressing electrographic seizure activity, but MK-801 had a slightly wider window for the prevention of chronic epilepsy after SE and early treatment, rather than electrographic suppression of SE, correlated with the prevention of chronic epilepsy [40]. The authors concluded, ''that studies of SE treatment should examine the effect of therapy on SE itself, as well as the long-term benefits of each treatment." NMDA antagonists, such as MK-801, or ketamine ''should be considered early in the treatment of SE".
Interestingly, phenobarbital seems to have a supra-additive effect when used simultaneously with other ASMs. Claude Wasterlaiń s group investigated the combination of phenobarbital, ketamine, and midazolam in soman-induced SE in rats [41]. They used rats, which had implanted transmitters to record EEG, and exposed them to soman. The rats were treated with atropine sulfate and oxime HI-6 dichloride salt one min after exposure and with ASMs 40 minutes after the onset of SE. A triple therapy combination of phenobarbital, midazolam, and ketamine effectively reduced time in SE, prevented soman-induced epileptogenesis, and reduced neurodegeneration. Phenobarbital in combinatorial treatment with midazolam and ketamine was more effective, that either substance alone in this cholinergic SE model [41]. Although dual therapy of phenobarbital with midazolam provided some neuroprotection in the piriform cortex, triple therapy with phenobarbital, ketamine, and midazolam protected additional brain regions from excitatory and inhibitory neuronal loss [41]. This further supports the potential of phenobarbital to be used in the earlier stages of SE in combination with other drugs. Currently, no clinical trial is on the horizon, to follow the concept of early combinatorial treatments [42] (see Table 2).

Clinical evidence for the use of phenobarbital in Status epilepticus
Despite its longstanding use in clinical practice, there are remarkably few randomized clinical studies on phenobarbital in SE [43][44][45][46], which will be reviewed here in brief.
In a randomized controlled, but non-blinded study, 36 consecutive patients with generalized convulsive SE (defined as SE lasting >30 min) were treated with either a combination of diazepam and phenytoin or with phenobarbital [43]. One group received diazepam at 2 mg/min IV until the patient stopped convulsing or after the 20 mg cumulative dose was reached. Phenytoin was administered simultaneously at a rate of 40 mg/min using a loading dose of 18 mg/kg. The other group received phenobarbital intravenously at 100 mg/min until a dose of 10 mg/kg was reached. Phenytoin was added as a rescue medication if SE persisted for 10 minutes after beginning therapy with phenobarbital. The main outcome Table 1 Key pharmacological characteristics of phenobarbital. measures were time to treatment response and cumulative convulsion time. With phenobarbital 16 of 18 patients (89%) had a clinical convulsion time of fewer than 10 minutes, and no patient remained in clinical convulsing SE for longer than 25 minutes. In the diazepam/phenytoin group only 10 of the 18 patients (56%) convulsed for less than 10 minutes and 5 (28%) had a cumulative convulsion time of greater than 25 minutes. The median convulsion time in the diazepam/phenytoin group was significantly longer than that with phenobarbital (9 minutes vs. 5 minutes; p < 0.06, two-tailed Wilcoxon rank test). The median time to response was significantly shorter with phenobarbital, compared to diazepam/phenytoin (5.5 minutes vs. 15 minutes; p < 0.1). The 18 patients in the phenobarbital arm received 5 to 23 mg/ kg and all but 2 patients required less than 12 mg/kg. The mean serum phenobarbital level in the responder was 18.3 lg/ml (range 15 to 27 lg/ml), which is lower than in chronic epilepsy treatment.

Mechanism of action
Overall, SE subsided with phenobarbital in 16 of 18 patients (89%), however, five of these patients ultimately received phenytoin for presumed additional clinical efficacy although no additional seizures were documented. Six patients (33%) in both groups had to be intubated; the rate of arrhythmias and hypotension was low and similar in both groups. The two protocols also differ in their practical use: While phenobarbital can be given as monotherapy through a single IV line, diazepam, and phenytoin need two different IV lines, which may lead to significant delays to achieve therapeutic serum levels. The authors concluded that ''the phenobarbital regimen appears at least as effective, comparable in safety, and enjoys practical advantages in comparison with the diazepam/phenytoin regimen" [43].
In the Veterans Affairs Status Epilepticus Cooperative Study, 570 patients with convulsive SE were randomized to four different treatments between 1990 and 1995 [44]. The doses used, were phenobarbital 15 mg/kg, lorazepam 0.1 mg/kg, phenytoin 18 mg/kg, and diazepam 0.15 mg/kg followed by phenytoin 18 mg/kg. The primary outcome was control of all visible motor and EEG seizure activity within 20 minutes after the start of treatment and no return of SE within the next 40 minutes. SE was defined as two or more generalized convulsions, without full recovery of consciousness between seizures, or continuous convulsive activity for more than 10 minutes. Patients were stratified as overt SE (defined as easily visible generalized convulsions) and subtle generalized convulsive status epilepticus, which was defined as coma and ictal EEG discharges with or without subtle convulsive movements, such as rhythmic twitching of the arms, legs, trunk, or facial muscles, tonic eye deviation, or nystagmoid eye jerking. Overt SE was stopped in 58.2% with phenobarbital, and 64.9% with lorazepam, which was statistically not significant. Both treatments were significantly better than phenytoin alone, which controlled only 43.6% (p = 0.002 and 0.006) [47]. The phenobarbital dose was 14.96 mg (±2.53 SD) yielding a serum level of 31.2 lg/ml (±37.2 SD); the mean infusion duration was 16.6 minutes (±11.5 SD). Approximately 50% of the patients received pretreatment, usually with diazepam, before enrollment. These patients were still eligible if SE was still ongoing. In this subgroup of pretreated patients, phenobarbital was the most successful agent [47]. The most important adverse effects of phenobarbital in overt SE were hypoventilation (13.2%), hypotension (34.1%), and cardiac arrhythmias (3.3%). However, these rates were in the same range as with lorazepam (10.3%, 25.8%, 7.2% respectively), diazepam/phenytoin (16.8%, 31.6%, 2.1%), and phenytoin (9.9%, 27.0%, 6.9%). The overall adverse event rate was higher in the subtle SE group, but again, did not differ between the treatment arms. Despite the fact, that there was no difference between phenobarbital and lorazepam the authors concluded ''As initial intravenous treatment for overt generalized convulsive status epilepticus, lorazepam is more effective than phenytoin. Although lorazepam is no more efficacious than phenobarbital or diazepam and phenytoin, it is easier to use." [44]. This conclusion seems hard to justify, as both, lorazepam and phenobarbital are used as a single-line injection, as compared to the combination of diazepam and phenytoin, which needs two IV lines.
Another randomized controlled trial from China enrolled 73 adult patients with benzodiazepine-resistant SE between February 2011 to August 2015 [45]. SE was operationally defined as 5 min or more of continuous clinical and/or electrographic seizure activity or recurrent seizure activity without recovery between seizures. The main etiology in this study was viral encephalitis 41.7% (30/72). Initially, all patients received 0.2 mg/kg IV diazepam, administered twice at a 10-min interval in case SE was not stopped. In the phenobarbital group, a loading dose of 20 mg/kg (with the option of an additional 5-10 mg/kg) at a rate of 50 mg/ min was applied, followed by an IV dose of 100 mg every 6 h. In  [45]. The same group conducted another study with the same design, but with some technical improvements, such as uniform intravenous pumping, pump speed adjustment according to adverse events, blood drug level monitoring, and continuous EEG monitoring [46]. In this second (''China 2-P vs. V") multicenter study 69 patients were enrolled. The rate of SE termination (<1 h) was significantly higher in the phenobarbital group (33 cases) than in the valproate group (36 cases) (84.8 % vs. 63.9 %, p = 0.048), but astonishingly the rates of nontermination of electrographic status within one hour were similar between the two groups (12.1 % vs. 8.3 %, p = 0.702). In the China 2-P vs. V study, there were no differences in the adverse event rate and the relapse rate between the groups [46]. In the Special Issue ''Proceedings of the 7th London-Innsbruck Colloquium on Status Epilepticus and Acute Seizures'' in this Journal, Brigo et al. published a systematic review and network metaanalysis on five randomized controlled trials in 349 adults with benzodiazepine-resistant convulsive SE [48]. The interventions, which entered the network-meta-analyses were: intravenous valproate (20-30 mg/kg), phenytoin (20 mg/kg), diazepam (0.2 mg/ kg, then 4 mg/h), phenobarbital (20 mg/kg, then 100 mg every 6 h), lacosamide (400 mg), and levetiracetam (20 mg/kg). The primary efficacy outcomes were SE cessation within 1 h from drug administration and seizure freedom at 24 h. Safety outcomes were respiratory depression and hypotension. The authors performed pairwise meta-analyses for all outcomes, using a fixed effects model and then a network meta-analyses within a frequentist framework assuming equal heterogeneity parameter s across all comparisons [49]. The distributions of the potential effect modifiers, for instance, study and patient-level covariates, were balanced across all pairwise comparisons to allow for the networkmeta-analyses [50]. All studies included patients with convulsive SE, but etiology and SE duration differed remarkably across studies. The clinical definitions of SE and dosages of benzodiazepines as first-line treatment were homogeneous between different studies.
In this network meta-analysis phenobarbital was superior to phenytoin, valproate, diazepam, levetiracetam, and lacosamide with respect to SE cessation and performed better than valproate, diazepam, and lacosamide in the achievement of seizure freedom at 24 h (Figs. 3 and 4). Significantly different was the higher proportions of SE cessation (OR: 5.36; 95% CI: 1.87-15.36; p = 0.002) and seizure freedom at 24 h with phenobarbital, as compared to valproate (OR: 7.07; 95% CI: 2.52-19.86) [48].

Adverse effects of phenobarbital in Status epilepticus
The main concern about using phenobarbital is the dose-related CNS depressant effect, which is common to all barbiturates. Serum levels above 70 lg/ml almost invariably impair consciousness, which may contribute to postictal coma. In patients, who are already compromised it may add an additional burden of treatment to the burden of the underlying cause of SE, with the risk of decompensation [51]. Of more concern seems to be respiratory depression, hypotension, and the negative inotropic effect on the myocardium at high doses. Respiratory depression is most pronounced within the first hour after administration. It is a rather slow process and respirations, blood pressure, and pulse should be monitored, either continuously or at least every five minutes [47]. In the randomized controlled trials, there was no difference in hypotension, respiratory depression, and cardiac arrhythmias [43,44], but there was a clear trend towards more respiratory depression in the network metanalyses and the randomized controlled trial from China [45,46,48]. Very high doses of barbiturates have been used in children [52] and adults [53,54] with refractory SE. Maximum serum levels ranging from 70 to 344 mg/L were reported in a retrospective series of 50 children. Forty of them were intubated prior to high-dose phenobarbital treatment but recovered respiratory drive and could be removed from the ventilator despite these extraordinarily high serum levels [52]. In another retrospective series in adults, phenobarbital doses ranged between 40-140 mg/kg/day and achieved serum levels of 35.29 to 218.34 lg/ml without significant adverse events reported in the abstract [53]. Very high doses of phenobarbital were reported in a retrospective case series of 10 patients with super-refractory SE [54]. These severely compromised patients had a median duration of SE of 17.5 days (range 6-60) and received anesthetics for a median of 14.0 days (range: 2-54) before treatment with high dose phenobarbital achieving median serum levels of 151.5 lg/ml. Aside from respiratory depression and hypotension in 7 patients, the main concern was a systemic infection, which occurred in all patients including pneumonia (5/10), urinary tract infection (4/10), fungal infection (2/10), and line infection (2/10), either alone or in combination. Three of the patients had sepsis following a fungal infection or bacterial pneumonia. As most of the patients received multiple drugs and had viral encephalitis as a cause of SE, it is hard to draw the final conclusion of a causal relationship to the high dose phenobarbital treatment, but it seems likely that, similar to thiopental/pentobarbital, a strong immunosuppressant effect comes also with long-standing and high dose phenobarbital treatment [55][56][57].
Absolute contraindications are rare and include a history of acute intermittent porphyria [58] and known hypersensitivity to barbiturates [59].

Current place of phenobarbital in the treatment of Status epilepticus
After its market introduction in the early 20th century, phenobarbital has been produced in large amounts and its clinical utility in SE has been repeatedly emphasized. Janz suggested, based on previous publications [60,61] doses between 300 and 400 mg subcutaneously, intramuscularly, or intravenously as long as it is given early in the SE [62,63] if SE persists, another 400 mg may be given. In the absence of clinical trials until the late 1980ies, the most common doses were 10 mg/kg (usual adult dose 600-800 mg). As a deliverable of the first London-Innsbruck Colloquium on Sta-tus epilepticus in 2007 [64], a consensus-based treatment protocol for SE was published, which recommended IV phenobarbital as one of four treatment options for established SE (10-20 mg/kg at 100 mg/min) [65]. In the EFNS evidence-based guidelines, phenobarbital was not recommended at all [66]. Unfortunately, phenobarbital was not included in the ESET Trial [67] and therefore the evidence-based guidelines of the American Epilepsy Society Guidelines [68] and the Guidelines of the German, Austrian, and Swiss Neurological Societies [69], phenobarbital are listed only as an alternative (15 mg/kg at 100 mg/min), when levetiracetam, valproate or phenytoin/fosphenytoin is not available. In the Finish Current Care Guideline on pharmacotherapy of SE, phenobarbital is not even mentioned as an alternative option [70]. Despite, the disappearance of evidence-based guidelines, phenobarbital is still widely in use in low-to-middle-income countries worldwide.
There is a remarkable recent trend to use subcutaneous phenobarbital (off-label) for the treatment of agitation, seizures, and status epilepticus in the palliative care setting [71][72][73], as it has been done in the early days, but clinical trials in this emerging field are still missing.

Conclusion
Phenobarbital is a highly effective drug with GABA-ergic and anti-glutamatergic activity. Hence its use in early and established SE, but also refractory SE seems to be justified from a mechanistic point of view. There is a long-standing clinical experience over the past 50 years, and it is still one of the most effective drugs in SE. It is overall well tolerated and has several clinical advantages, including rapid and long-lasting action, and a favorable safety profile even at high doses. It is one of the few drugs for SE, where ample data are available in young children and newborns. However, there are remarkably few randomized clinical trials including Phenobarbital, which led to a downgrading in strictly evidence-based guidelines, emphasizing the need for further studies using Phenobarbital as a comparator drug in SE.

Consent for publication
Not applicable.