Seizure incidence during the neonatal period has been estimated to be higher than any other period in the lifespan, yet this is the period that we know the least about in terms of either the deleterious effect of seizures or of the drugs used in their treatment. Epilepsy poses a special challenge for developmental regulation because both the disorder and many of the current drug therapies used to treat it are at odds with normal maturational processes. Experimental and clinical evidence indicates that recurrent seizures in the immature brain can have long-lasting adverse consequences (Farwell et al., 1985; Huang et al., 1999; Schmid et al., 1999; Baram, 2003). On the other hand, recent experimental data also raise concerns about the potential detrimental impact of antiepileptic drugs (AEDs) during critical maturational periods (Bittigau et al., 2002; Olney, 2002; Glier et al., 2004; Olney et al., 2004b). Although the exact mechanisms responsible for these adverse outcomes are not well understood, one possible contributor can be the impairment of naturally occurring programmed cell death (PCD) caused by changes in the patterning of neural activity at critical developmental stages. Because epileptic seizures represent a state of hypersynchrony, while AEDs can impede activity-dependent processes, both influences can adversely challenge the developing brain during the last trimester of gestation and early infancy.
Summary: Seizure incidence during the neonatal period is higher than any other period in the lifespan, yet we know little about this period in terms of the effect of seizures or of the drugs used in their treatment. The fact that several antiepileptic drugs (AEDs) induce pronounced apoptotic neuronal death in specific regions of the immature brain prompts a search for AEDs that may be devoid of this action. Furthermore, there is a clear need to find out if a history of seizures alters the proapoptotic action of the AEDs. Our studies are aimed at both of these issues. Phenytoin, valproate, phenobarbital, and MK801 each induced substantial regionally specific cell death, whereas levetiracetam even in high doses (up to 1,500 mg/kg) did not have this action. In view of our previously findings of neuroprotective actions of repeated seizures in the adult brain, we also examined repeated seizures for a possible antiapoptotic action in the infant rat. Rat pups were preexposed to electroshock seizures (ECS) for 3 days (age 5–7 days) before receiving MK801 on day 7. The effect of ECS, which was consistently a 30% decrease in MK801-induced programmed cell death (PCD), suggests that repeated seizures can exert an antiapoptotic action in the infant brain. In contrast, PCD induced by valproate was not attenuated by ECS preexposure, suggesting that valproate-induced PCD is mechanistically distinct from that induced by MK801 and may not be activity-dependent. Presently, we do not know if this neuroprotective effect of seizures is deleterious or beneficial. If the seizures also enhance the survival of neurons that are destined to undergo naturally occurring PCD, early childhood seizures may have deleterious effects by preventing this necessary component of normal development. While this effect of seizures might be counteracted by AEDs, the fact that several AEDs shift the PCD to the other extreme, and trigger excessive neuronal cell loss, raises concern about whether the drug therapy may be more detrimental than the seizures. In this context, it is encouraging that we have identified at least one AED that is devoid of a proapoptotic action in the infant brain, even in high doses. It is now important to evaluate the long-term consequences of the changes in PCD in infancy by examining behavioral outcomes and seizure susceptibility in the AED- and seizure-exposed neonates when they reach adulthood.
AED-induced cell death in the neonatal brain
Most traditional AEDs induce widespread apoptotic neuronal death in specific regions of the neonatal brain (Bittigau et al., 2002). This effect resembles that produced by ethanol, anesthetics, and the glutamate antagonist, MK801 (Ikonomidou et al., 1999; Olney et al., 2004a). Moreover, the combination of these drugs can cause profound cell death, even when given in doses that are subthreshold for inducing cell death with individual drug treatments (Bittigau et al., 2002; Jevtovic-Todorovic et al., 2003). The developmental period when AEDs exert their proapoptotic activity coincides with a brain “growth spurt” period characterized by a high rate of synaptogenesis and developmental apoptosis (Bittigau et al., 2002). In humans, this period stretches from the third trimester of pregnancy through early infancy (Bittigau et al., 2002), corresponding to the first 10 days postnatally in rats.
Because exposure of neonatal rats to proapoptotic drugs has been associated with long-term adverse behavioral outcomes (Harris et al., 2003; Jevtovic-Todorovic et al., 2003), it is possible that proapoptotic actions contribute to these outcomes in rats and to the cognitive impairment observed in humans following exposure to certain AEDs in utero or early postnatal development (Reinisch et al., 1995; Adab et al., 2004). In the interest of minimizing the risk of long-term adverse effects of seizure treatment, the identification of AEDs devoid of a proapoptotic action is a top priority. Newer AEDs having mechanisms of action distinct from those of traditional AEDs may offer the possibility of avoiding this deleterious effect. Levetiracetam (Keppra), is a newer AED possessing a unique mechanism of action (Klitgaard, 2001). Therefore, we have conducted studies to examine the effect of levetiracetam on cell death in several brain regions in neonatal rat pups and compared it with MK801, phenytoin, and valproate, which have previously been shown to have proapoptotic effects (Ikonomidou et al., 1999; Bittigau et al., 2002). In addition, we examined whether levetiracetam influenced the cell death induced by MK801 or phenytoin.
In the adult rat brain, repeated brief electroconvulsive shock (ECS) seizures exert profound neuroprotective effects (Masco et al., 1999; Kondratyev et al., 2001). However, the same seizure-induced “preconditioning” that may be advantageous in the brain of the adult animal (e.g., to guard against further injury) may be maladaptive in the brains of immature animals. One can imagine that repeated seizure activity in immature animals could enhance the survival of neurons that are destined to undergo naturally occurring PCD during early postnatal development. Thus, one reason that early childhood seizures may have deleterious effects is not because they can cause cell loss, but because they might actually prevent the cell loss that is a necessary component of normal development. Presumably, this effect of seizures could be counteracted by treatment with appropriate AEDs, or, conversely, a prior history of seizures may attenuate the drug-induced PCD in the neonate. Therefore, we have also examined the impact of repeated seizures on drug-induced PCD in the immature brain.
Animals and drug treatment for testing acute neurotoxicity of drugs
Postnatal day (PD) 7–8 Sprague–Dawley rat pups (Harlan) were used. MK801 (0.5 mg/kg, Sigma), phenytoin (sodium diphenylhydantoin, 50 mg/kg, Sigma), and sodium valproate (400 mg/kg, Sigma) were injected intraperitoneally (i.p.) on PD 7, the age of peak vulnerability to drug-induced cell death. Drugs were dissolved in saline except for phenytoin, which was in alkalinized saline (pH 9–11). Levetiracetam (Keppra oral solution, UCB Pharma, Smyrna, GA, U.S.A.) was administered i.p. (250, 1,000, or 1,500 mg/kg), or orally (500 mg/kg) on PD 7. A total of 1,500 mg/kg of levetiracetam was given in divided doses: 1,000 mg/kg at T = 0 h and 500 mg/kg at T = 8 h. Control groups received normal or alkalinized saline. Brain samples were collected on PD 8 at T = 24 h. For drug combination studies, levetiracetam was given as above and MK801 or phenytoin was given at T = 2 h and animals were sacrificed at T = 26 h.
Measurement of levetiracetam concentration
Whole brain was collected and homogenized in 4 volumes of phosphate buffered saline at T = 2 h after levetiracetam administration (500 mg/kg, p.o. or 1,000 mg/kg, i.p.). Drug concentration was determined by HPLC as previously described (Pucci et al., 2004), and the identity of levetiracetam from the HPLC elution was confirmed by mass spectrometry analysis.
The combination of repeated brief electroconvulsive shock seizures with drug treatment
Minimal ECS was administered in a standard fashion via corneal electrodes (60 Hz, 200 ms, 35 mA) delivered by a Wahlquist stimulator (Wahlquist Instrument Company, Salt Lake City, UT, U.S.A.) as described previously (Kondratyev et al., 2001). Control (sham) animals received the same handling and contact with the electrodes, but no current was passed. Animals were behaviorally observed to ensure that minimal motor seizures (head bobbing, pedaling, clonic movements of forelimbs) lasting more than 5–10 s occurred after each ECS. A single daily ECS treatment session consisted of three ECS seizures, given at 30-min intervals (i.e., at 0, 30, and 60 min). Repeated ECS consisted of daily treatment for 3 days given from PD 5 to PD 7. In order to study the effect of preexposure to seizures on drug-induced neurodegeneration, MK801 (0.5 mg/kg, 3 consecutive i.p. injections at T = 0, 8, and 16 h) or valproate (400 mg/kg, i.p.) was administered 1 h after the last ECS treatment on PD7. Animals were sacrificed on PD 8, 24 h after the first drug injection to measure the extent of cell death.
Histology and quantification
Coronal cryostat sections (20 μm) were prepared from quick frozen brain. To measure apoptotic cell death, TUNEL staining was performed using the Apoptag peroxidase in situ apoptosis detection kit (Chemicon International, Inc., Temecula, CA, U.S.A.). Staining was examined in photomicrographs (10×) of three sequential sections at 200 μm intervals taken from several brain areas; quantification of cell death was performed in three brain areas which exhibited the greatest degeneration across treatments (ventral thalamus, dorsomedial anterior striatum, and cortex [frontal or retrosplenial]). TUNEL-positive cells within 1.00 mm2 were counted regardless of size and shape by an observer blind to the treatment conditions. Statistical comparisons were performed by analysis of variance (ANOVA) followed by post hoc tests.
Fluoro-Jade B staining was performed as described previously (Schmued and Hopkins, 2000) and sections were examined using a fluorescent microscope (Nikon Eclipse E800 Nikon Instruments Inc., Melville, NY, U.S.A.) with filters for FITC.
Drug treatment effects on behavior, brain, and body weight in neonatal rats
MK801 and valproate caused sedation and impaired body weight gain. Compared to controls, MK801-treated pups had reduced brain (9.5%) and body weights (18.6%), resulting in an 11% increase in the brain-to-body weight ratio. Valproate-treated pups had reduced brain weight (8%) as previously reported (Bittigau et al., 2002); with the reduction in body weight (16%), there was a 9.4% increase in the brain-to-body weight ratio. Phenytoin had significantly less impact on brain and body weight gain than MK801 or valproate. In contrast, levetiracetam caused neither a change in behavior nor in growth (Table 1).
|Group (n)||Brain weight (mg)||Body weight (g)||Δ body weight for 24 h||Brain/body weight ratio × 100|
|Control (11)||820.0 ± 11.5||17.66 ± 0.61||2.37 ± 0.13||4.69 ± 0.12|
|Valproate (8)c||754.6 ± 8.13*||14.83 ± 0.56||−0.20 ± 0.43*||5.13 ± 0.17|
|Levetiracetam 250 (4)||762.0 ± 21.2||16.09 ± 0.54||1.85 ± 0.16||4.86 ± 0.11|
|Levetiracetam 500 (6)||806.1 ± 9.40||17.05 ± 0.45||2.38 ± 0.08||4.73 ± 0.03|
|Levetiracetam 1,000 (4)||830.0 ± 27.4||18.10 ± 0.80||1.85 ± 0.25||4.60 ± 0.05|
|Levetiracetam 1,500 (7)||790.0 ± 3.09||17.64 ± 0.85||2.47 ± 0.26||4.48 ± 0.22|
|Phenytoin (11)a||790.9 ± 11.2||17.13 ± 0.55||1.27 ± 0.19*||4.70 ± 0.10|
|Phenytoin + levetiracetam 500 (8)b||752.1 ± 8.34*||16.17 ± 0.67||0.91 ± 0.24*||4.67 ± 0.18|
|MK801 (11)||742.5 ± 10.1*||14.38 ± 0.47*||−0.05 ± 0.25*||5.21 ± 0.15|
|MK801+ levetiracetam 500 (6)b||727.5 ± 19.7*||15.32 ± 0.72||−0.02 ± 0.18*||4.79 ± 0.20|
|MK801 + levetiracetam 1,500 (9)b||732.2 ± 8.63*||14.23 ± 0.82*||−0.11 ± 0.20*||5.29 ± 0.33|
Levetiracetam: effects on cell death in 7-day-old rat pups
The extent of cell death 24 h following levetiracetam was compared with that of phenytoin, MK801, and valproate. As previously reported (Ikonomidou et al., 1999; Bittigau et al., 2002), valproate, phenytoin, and MK801 significantly increased the number of degenerating cells in multiple brain regions (Fig. 1). However, levetiracetam, even in very high doses, did not increase TUNEL positive cells in any of the brain regions (Fig. 1). Moreover, repeated doses of levetiractam (1,000 mg/kg followed by 500 mg/kg after 8 h) did not promote cell death (Fig. 1). Fluoro-Jade B staining profiles were consistent with TUNEL assay results (Fig. 2). When measured at 2 h after treatment, high levels of levetiracetam were found in the brain; at 2 h after 500 mg/kg oral, the level was 1,265.30 ± 350.15 μmol/L; after 1,000 mg/kg i.p., the level was 4,517.35 ± 348.60 μmol/L (n = 3). These concentrations correspond to levels well above those in the therapeutic range (35–120 μmol/L) in human (Johannessen et al., 2003).
Levetiracetam: effect on cell death induced by MK801 or phenytoin
In order to determine whether levetiracetam exacerbates the cell death produced by proapoptotic drugs, rat pups were given a combined regimen of levetiracetam + MK801 or phenytoin. A total of 500 mg/kg of levetiracetam neither exacerbated nor reduced the extent of cell death induced by MK801 or phenytoin (Fig. 1). Although an exceptionally high dose (1,500 mg/kg) of levetiracetam had no effect on MK801-induced cell death in striatum and frontal cortex, a modest increase in cell death was detected in ventral thalamus (Fig. 1).
Repeated brief seizures: effects on drug-induced cell death in 7-day-old rat pups
Preexposure of rat pups to 3 days of ECS treatment, between postnatal days 5 and 7, significantly attenuated the cell death induced by MK801 in striatum and thalamus (Figs. 3 and 4) but not in retrosplenial cortex. In contrast, neuronal cell death induced by valproate was not attenuated by the ECS preexposure. The ECS treatment by itself did not cause an increase in cell death in any of the brain areas examined.
Our results demonstrate that exposure of neonatal rats to seizures as well as several traditional AEDs can affect the extent of PCD, a vital feature of early postnatal neural development, in the immature rat brain. First, we observed that, in contrast to several traditional AEDs (Bittigau et al., 2002), levetiracetam does not induce cell death in the developing rat brain even in doses several fold higher than therapeutic doses. This extends the findings of Manthey et al. (2005) who found no proapoptotic effect of lower doses of levetiracetam (5–100 mg/kg) in neonatal Wistar rats. Additionally, repeated doses (1,000 mg/kg followed by 500 mg/kg after 8 h) did not result in detectable cell death, ensuring that this toxicity is not evident even with more extended exposure to the drug. Moreover, unlike other drugs examined, levetiracetam induced no behavioral side effects or growth retardation. In contrast, we observed that although phenytoin had relatively little effect on behavior, body, or brain weight, it caused substantial neuronal apoptosis in the neonatal rats. This suggests that the cell death induction can occur even in the absence of significant sedation or growth retardation. Moreover, growth retardation per se does not account for the cell death response, since exposure of control groups to nutritional and maternal deprivation equivalent to that of drug-treated pups did not result in cell death (Olney et al., 2004a).
Levetiracetam did not exacerbate cell death induced by MK801 or phenytoin even when given at a dose (500 mg/kg) that produced drug levels well above the therapeutic range. This contrasts with findings with combinations of other AEDs, which induce supra-additive toxicity at low doses (Bittigau et al., 2002). Moreover, our recent experimental data indicate that despite the relative safety of several newer AEDs when given alone, many of these drugs exert proapoptotic effects when given in combinations (Katz et al., 2007; Kim et al., 2007). The results with 1,500 mg/kg levetiracetam suggest that the potential for enhanced toxicity to other drugs exists only in the event of an extreme overdose of leviteracetam (30-fold higher than therapeutic). Thus, using this parameter, levetiracetam appears to have a better safety profile than traditional AEDs both alone and in polytherapy. Based on our recent findings with carbamazepine (Kim et al., 2007), it appears that cell death is also avoided when carbamazepine is given alone or in combination with levetiracetam, but not when it is given in combination with phenytoin.
Levetiracetam has many potential advantages as a treatment in early childhood. It is a broad-spectrum AED having high oral bioavailability, linear kinetics, and almost no drug interaction with an excellent safety profile (Perucca and Johannessen, 2003). Moreover, it appears to be well tolerated due to a relatively low incidence of adverse effects (Wheless and Ng, 2002; Koukkari and Guarino, 2004; Grosso et al., 2005; Lagae et al., 2005). While the dosage range of levetiracetam is often 10–60 mg/kg/day in children (Wheless and Ng, 2002; Grosso et al., 2005; Lagae et al., 2005), higher doses have been also used (Wheless and Ng, 2002; Koukkari and Guarino, 2004). Clinical trials in children including infants have shown that levetiracetam was effective in partial and generalized seizures both in monotherapy and polytherapy (Wheless and Ng, 2002; Grosso et al., 2005; Lagae et al., 2005). In fact, this drug may have a better efficacy and safety profile in infants and children less than 4 years old compared with older children (Grosso et al., 2005). A case study with three pregnant women with epilepsy has documented that levetiracetam monotherapy during pregnancy did not cause any adverse effects in either mothers or infants, while controlling seizures effectively (Long, 2003).
Levetiracetam has been proposed to have neuroprotective effects (Hanon and Klitgaard, 2001; Marini et al., 2004); however, it did not prevent the apoptotic effect of MK801 or phenytoin in our studies. This lack of antiapoptotic action could be viewed as advantageous in the developing brain, because apoptotic PCD is an essential component of normal brain maturation. Based on this benign profile and the favorable pharmacokinetics and other safety features, levetiracetam may be an especially good candidate for both monotherapy and polytherapy in neonates and women during late pregnancy. However, our results so far are based on acute treatments; studies with chronic treatment in immature animals are needed before making conclusions about the safety of this drug for AED therapy in infants or pregnant women.
Repeated brief seizures have been found to have profound neuroprotective effects in the adult brain (Masco et al., 1999; Kondratyev et al., 2001). This led us to ask the question of whether brief seizures may attenuate the drug-induced PCD in the immature brain. Our results indicate that the preexposure to repeated brief electroshock seizures significantly attenuated the MK801-induced degenerative response in thalamus and striatum. This effect of ECS, which was consistently a 30% decrease in PCD, suggests that repeated seizures can exert an antiapoptotic action in the infant brain. However, this antiapoptotic effect of seizures was not observed in the retrosplenial cortex, raising the possibility that the mechanism of induction of cell death in this region may be distinct from that in striatum and thalamus.
The fact that ECS preexposure did not attenuate the PCD induced in striatum and thalamus by valproate (400 mg/kg) suggests that the mechanism of induction of PCD by valproate is likely to be distinct from that of MK801. This raises the possibility that the valproate-induced neuronal death may not be activity-dependent, and therefore insensitive to attenuation by ECS. Valproate may promote apoptosis by acting as a histone deacetylase inhibitor, an effect that may also contribute to the teratogenicity of this drug (Eikel et al., 2006). This may also explain why valproate can enhance PCD in the neonatal rat brain even when given in doses that are well below the range for antiepileptic effects (Bittigau et al., 2002). Additional evidence that MK801 and valproate cause PCD through distinct mechanisms comes from comparing the regional pattern of cell death caused by these two drugs. To make such a comparison, multiple doses of MK801 are needed to compensate for the short half-life of the drug (2 h) (Ikonomidou et al., 1999) relative to that of valproate (9 h in P7 rats) (Mares et al., 1989). As evident in Fig. 3, we observed that whereas MK801 (given in 3 doses at 8 h intervals) causes at least as much PCD in striatum as in thalamus, the effect of valproate on striatum is almost 3-fold lower than its effect on thalamus, even though the effect on thalamus is slightly greater than that of MK801. Thus, it appears that the thalamic neurons are considerably more sensitive than the striatal neurons to the toxic action of valproate, a dissociation not observed with MK801.
Our finding that brief seizures attenuated the MK801-induced PCD may have implications for the impact of repeated seizures on the immature brain. On the one hand, if we view MK801 exposure as a model for deprivation-induced PCD (as might occur with sensory deprivation, hypoxia, ischemia, or other localized neural injury), then the counteracting effect of the brief seizures could be viewed as adaptive or compensatory. On the other hand, if we view MK801 treatment as a tool to amplify the naturally occurring PCD, then the effect of seizures could be viewed as interfering with a necessary component of development. To evaluate the possibility that seizures may interfere with naturally occurring PCD in the neonatal rat brain, baseline PCD would need to be measured throughout an entire brain region, not just in samples of a region as we have done in the present study. The sampling technique used in the present study is quite sensitive to increases in PCD above the constitutive level (which is very low compared to the level induced by drug treatment), but it is insensitive to detecting a decreased level of PCD against the low baseline. Thus, additional studies are warranted to determine whether repeated seizure episodes disrupt native developmental PCD.
It will be important to determine whether long-term outcomes of exposure to repeated seizures and/or AEDs are dependent on the severity of the changes in PCD that occur during this highly defined critical period in development. If so, then we should be able to define the critical time window and treatment durations that predispose to undesirable long-term effects, and identify strategies for avoiding these consequences. An important goal is to determine the conditions under which therapy during pregnancy, in preterm infants, or in neonates, may favor adaptive outcomes while minimizing deleterious consequences for development. In this context, it is encouraging that we have been able to identify at least one AED that is devoid of a proapoptotic action in the developing brain, even in high doses. The fact that seizures may exert an antiapoptotic action underscores the importance of evaluating the effects of AEDs against a background of seizure episodes. Future studies are needed to evaluate the long-term functional consequences of the changes in PCD during development, by examining behavioral outcomes and seizure susceptibility in the AED- and seizure-exposed neonates when they reach adulthood.
Acknowledgments: This study was supported by a Predoctoral Fellowship from the Epilepsy Foundation, and a research grant from the Partnership for Pediatric Epilepsy Research, administered through the Epilepsy Foundation, and NIH grants NS 20576, MH 02040, U10 HD047890. We thank Samantha Crowe for helpful discussions, and Hatice Ozel Abaan (Lombardi Comprehensive Cancer Center) and Dr. Yoshiyuki Tokiwa (US Food and Drug Administration) for assisting with the measurement of drug concentrations.