Rational Polypharmacy: When Two Old Drugs Are Better than One

Abstract

Bumetanide Enhances Phenobarbital Efficacy in a Neonatal Seizure Model. Dzhala VI, Brumback AC, Staley KJ. Ann Neurol 2008;63(2):222–235. OBJECTIVES: High levels of expression of the Na⁺-K⁺-2Cl^- (NKCC1) cotransporter in immature neurons cause the accumulation of intracellular chloride and, therefore, a depolarized Cl^- equilibrium potential (E_Cl). This results in the outward flux of Cl^- through GABA_A channels, the opposite direction compared with mature neurons, in which GABA_A receptor activation is inhibitory because Cl^- flows into the cell. This outward flow of Cl^- in neonatal neurons is excitatory and contributes to a greater seizure propensity and poor electroencephalographic response to GABAergic anticonvulsants such as phenobarbital and benzodiazepines. Blocking the NKCC1 transporter with bumetanide prevents outward Cl^- flux and causes a more negative GABA equilibrium potential (E_GABA) in immature neurons. We therefore tested whether bumetanide enhances the anticonvulsant action of phenobarbital in the neonatal brain. METHODS: Recurrent seizures were induced in the intact hippocampal preparation in vitro by continuous 5-hour exposure to low-Mg²⁺ solution. The anticonvulsant efficacy of phenobarbital, bumetanide, and the combination of these drugs was studied. RESULTS: Phenobarbital failed to abolish or depress recurrent seizures in 70% of hippocampi. In contrast, phenobarbital in combination with bumetanide abolished seizures in 70% of hippocampi and significantly reduced the frequency, duration, and power of seizures in the remaining 30%. INTERPRETATION: Thus, alteration of Cl^- transport by bumetanide enables the anticonvulsant action of phenobarbital in immature brain. This is a mechanistic demonstration of rational anticonvulsant polypharmacy. The combination of these agents may comprise an effective therapy for early-life seizures.

Commentary

Although traditionally assigned the role as the main inhibitory influence that prevents excessive excitation and seizures in the brain, the neurotransmitter GABA plays a much more complex role than initially imagined. As reviewed in recent Epilepsy Currents commentaries (1,2), depending on the intracellular chloride content of postsynaptic neurons, GABA may also be excitatory, even promoting epileptiform activity. The postsynaptic effect of GABA stems from the opening of chloride channels within the GABA_A-receptor complex: when intracellular chloride is kept low because of its extrusion by the K⁺/Cl^–cotransporter isoform 2, KCC2, GABA_A-receptor activation causes the flow of negatively charged chloride ions into the cell. The resulting hyperpolarization of the neuronal membrane is the basis for the inhibitory effect on postsynaptic neurons and for the antiepileptic effects of drugs (e.g., benzodiazepines and barbiturates) that enhance GABAergic neurotransmission. However, when intracellular chloride is high, which occurs when KCC2 expression is reduced or when the Na⁺/K⁺/Cl^–cotransporter isoform 1 (NKCC1) is expressed, GABA_A-activated chloride flow is reversed, and GABA is depolarizing, even surpassing the threshold for action potential generation. Because the reversal potential for GABA_A-activated chloride currents is never as depolarized as that for the glutamate-receptor activated currents, GABA can still have a “shunting” effect on excitatory potentials and remain a relatively stabilizing influence on excitability.

During the early postnatal period in rodents and likely during the late prenatal and early postnatal period in humans, NKCC1 is expressed highly in the brain, while KCC2 expression is low. Under these conditions, GABA release contributes to the initiation of giant depolarizing potentials (GDPs) in populations of neurons. These GDPs not only serve as a marker of GABAergic excitatory influence but also play a critical role in the development of mature synapses. As the brain matures, NKCC1 expression is downregulated; however, recent evidence points to a recapitulation of this early developmental pattern of chloride transporter expression in neurons exposed to injury and in hippocampal neurons from patients with mesial temporal lobe epilepsy.

Previous studies have shown that the excitatory effect of GABA is at least partially responsible for the increased susceptibility to seizures in neonates (3). Bumetanide, a loop diuretic that potently inhibits NKCC1 transporter, impairs the intracellular accumulation of chloride and thereby, may convert GABA from excitatory to inhibitory. Accordingly, bumetanide abolishes GDPs in neonatal brain and prevents experimental seizures both in vitro and in vivo (3). Not surprisingly, because phenobarbital potentiates the effects of GABA, it is ineffective in reducing seizure activity in the same seizure model.

In clinical practice as well, phenobarbital fails to control acute seizures in more than half of neonates (4). Nevertheless, many physicians continue to choose barbiturates or benzodiazepines as first line agents, perhaps because of their familiarity and lack of compelling data for alternate therapies. In the present study by Dzhala et al., a possible new treatment strategy was tested in the established in vitro low-magnesium seizure model, using intact hippocampi from neonatal rats. This study design may provide a more robust model of hippocampal seizures because the hippocampal circuitry remains intact and the induction of seizures did not involve alteration of ion gradients that may influence chloride homeostasis. The authors hypothesized that if inhibition of NKCC1 with bumetanide alters the chloride gradient of neonatal hippocampal neurons such that GABA becomes more hyperpolarizing, then phenobarbital will gain the seizure suppressive effect that it has in mature brains.

To test their hypothesis, the authors first showed the effects of these drugs on chloride homeostasis and neuronal activity. The reversal potential for GABA_A-activated currents in immature pyramidal neurons was −63 mV, compared with a resting potential of −70 mV, indicating that GABA was in fact depolarizing. Consistent with this result, a GABA_A-receptor agonist (i.e., isoguvacine) exerted a net excitatory effect, increasing neuronal firing. As expected, although phenobarbital augmented GABA_A-activated currents, it did not reduce its excitatory effect. When bumetanide was added, the reversal potential for GABA shifted to −73 mV, suggesting that GABA had become inhibitory. Importantly, the administration of phenobarbital in combination with bumetanide now suppressed neuronal firing.

The authors next asked whether bumetanide could alter the effect of phenobarbital on seizure activity. Bathing intact hippocampi from neonatal rats in a low-magnesium solution induced recurrent seizure-like discharges, which was probably related to hyperexcitability induced when the magnesium block of the NMDA receptor was relieved. In this model, phenobarbital alone stopped seizure-like events in 30% of hippocampi and decreased frequency in others; however, it also increased duration and amplitude of the power spectra of the population activity in those hippocampi, indicating that by some measures the remaining seizures were worsened. Bumetanide alone had some efficacy against seizures: although it abolished seizures in only 20% of hippocampi, it decreased frequency, duration, and amplitude of power spectra in the others. The combination of bumetanide with phenobarbital produced a much stronger effect: seizures were abolished in 70% of hippocampi, and their frequency, duration, and power spectra were reduced in the remaining hippocampi more than with bumetanide alone. These findings support the clinical observation that phenobarbital alone is unlikely to stop seizures in neonates, and bumetanide, although itself weakly anticonvulsant, is also unlikely to produce a clinically satisfactory effect. However, the combination of increasing the inhibitory efficacy of GABA with bumetanide and potentiating its effect with phenobarbital, even when administered simultaneously, has the potential to be a powerful anticonvulsant therapy for neonatal seizures.

While Dzhala et al. used an animal model to demonstrate the efficacy of bumetanide and phenobarbital to decrease artificially induced seizures, the application of these two drugs to the treatment of human epilepsy is logical. Presumably, the strategy will work with benzodiazepines as well. The authors argue that bumetanide has demonstrated safety in neonates, and in vivo studies suggest effective penetration into brain (3); however, Dzhala and colleagues are careful to note the uncertain consequences of altering the delicate balance of inhibitory and excitatory influences in the developing brain using bumetanide. As previously mentioned, depolarizing GABA responses such as GDPs play an important role in the development of mature synapses, therefore long-term administration of bumetanide could have unintended and potentially untoward effects that may manifest later in life. Phenobarbital itself has been implicated in deleterious effects on neuron survival (5) and in persistent cognitive deficits after use during critical developmental stages (6). Therefore, greater efficacy of a drug combination against typically treatment-resistant neonatal seizures may allow for use of lower doses or shorter duration of phenobarbital therapy. In addition, the benefit of providing earlier control of seizure activity may itself reduce potential long-term consequences of neonatal seizures. Finally, because of mounting evidence for altered chloride homeostasis in adult epileptic brain, the prospective use of this combination therapy may extend beyond neonates to rational combinations of bumetanide and GABA-enhancing medications in treatment-resistant adult epilepsy.