Overall seizure recurrence
Our cohort’s seizure outcomes fall within the previously reported rates of seizure freedom following temporal lobe resections (Sperling et al., 1996; Foldvary et al., 2000; McIntosh et al., 2001; Salanova et al., 2002; McIntosh et al., 2004; Paglioli et al., 2004; Spencer et al., 2005; Jeha et al., 2006). In addition, as expected with TLE surgery (Fong et al., 2011), a significant proportion of our patients with breakthrough seizures regained seizure control and achieving an Engel I classification by last follow-up. This study also reproduces the traditional outcome predictors identified in prior TLE surgery reports, including higher rates of seizure freedom with unilateral MRI lesions and clear epileptic pathology (Yoon et al., 2003; McIntosh et al., 2004; Tonini et al., 2004; Tellez-Zenteno et al., 2005; Jeha et al., 2006), and worse seizure outcomes in patients with a history of generalized tonic–clonic seizures (Hennessy et al., 2001; McIntosh et al., 2004; Jeong et al., 2005; Spencer et al., 2005; Jeha et al., 2006) or frequent preoperative seizures (Foldvary et al., 2000; Jeha et al., 2006). Patients taking a higher number of AEDs at the time of surgery may have a more “severe” or “refractory” epilepsy, less amenable to a surgical cure. All these confirmatory findings do not augment the current literature, but emphasize that our cohort is indeed representative of a “typical” TLE surgery population.
Our main and novel finding is the improvement in seizure freedom following TLE surgery with the perioperative use of LEV. This favorable effect was unique to LEV in our cohort, similar whether LEV was used in monotherapy or polytherapy, and was in fact observed despite a higher proportion of several poor outcome determinants in the LEV cohort. The robust favorable LEV effect (adjusted p-value = 0.003) persisted after multivariate analysis controlling for currently known clinical and radiologic outcome predictors, prior and concomitant AED use, and year and type of resection suggesting that it is difficult to attribute it to improved diagnostic or surgical methods used in the LEV cohort. In fact, we evaluated the available data as best as possible within the limitations of a retrospective cohort to look for and adjust for possible confounders. As such, the possibility of a true physiologic basis for improved seizure outcomes with LEV deserves careful consideration.
One hypothesis might be that LEV is a better “antiepileptic” medication in postsurgical patients. In fact, one report has already suggested better control of recurrent postoperative seizures using LEV (Janszky et al., 2005). Although such an assumption is plausible, it would be difficult to speculate how surgery will selectively convert the LEV group in our intractable TLE cohort to pharmacoresponsiveness. Furthermore, although all AEDs were equally effective soon after surgery, with 75–86% of our patients being seizure free at four postoperative months (Table 4), the seizure freedom advantage for patients taking increased over time to reach up to a 20% difference for example compared with the carbamazepine (CBZ) group at 60 months. Discontinuation rates of the various AEDs were similar in our cohort (Fig. 1), so it is reasonable to speculate that adherence to AED treatment was also similar across groups. As such, a growing gap in the rates of seizure freedom for patients taking LEV introduces a postoperative time-dependency element to the LEV effect, which is difficult to explain through “anticonvulsant” properties alone.
An alternative hypothesis is that better seizure outcomes with LEV use might be related to an “antiepileptogenic” effect. Assuming that surgery removed the current focus of intractable epilepsy, improved effectiveness of LEV in the same patient population suggests that LEV may be acting mainly as an “antiepileptogenic” medication after surgery as opposed to its more typical “antiepileptic” role before the resection. A recent study of rats with genetic predisposition to epilepsy showed that LEV treatment starting prior to epilepsy onset decreased subsequent seizure frequency and duration (Yan et al., 2005). Proposed mechanisms of this antiepileptogenic effect in animals range from inhibition of interleukin-1β inflammatory markers in the hippocampus and piriform cortex or transforming growth factor β in astroglia, to modulation of presynaptic P/Q-type voltage-dependent calcium channels in granule cells of dentate gyrus (Lee et al., 2009; Kim et al., 2010; Stienen et al., 2011). Human brain tissue studies have also shown alteration of synaptic vesicle release within neurons and stabilization of epileptic γ-aminobutyric acid (GABA)A receptors (Palma et al., 2007; Yang & Rothman, 2009). However, direct evidence translating these experimental data into clear beneficial clinical effects has been lacking. One study found that adjunctive therapy with LEV is more effective when used early in patients who failed to respond to TLE surgery, than in refractory patients who had no prior surgery (Motamedi et al., 2003), but a subsequent retrospective analysis of temporal lobectomy patients did not show any superiority of potentially neuroprotective AEDs when compared with PHT or CBZ in preventing postoperative seizure recurrence (Asadi-Pooya et al., 2008). However, the study by AsadiPooya et al. was underpowered as it included a smaller group of patients taking “neuroprotective” AEDs (a total of 53 patients taking LEV, TOP, ZNS, and TGB combined), whereas our study includes 124 patients taking LEV alone. By the same token, smaller numbers of patients in our cohort taking the remaining potentially neuroprotective drugs (topiramate [TOP], zonisamide [ZNS], tiagabine [TGB]) may account for the lack of any favorable effects.
An underlying concept behind attributing better seizure outcomes to an antiepileptogenic effect of any medication though is the assumption that postoperative seizure recurrence is at least partly due to epileptogenesis. Our previous work on longitudinal postoperative seizure outcome supports this notion. Following resective surgery for temporal (Jeha et al., 2006), frontal (Jeha et al., 2007), or posterior quadrant (Jehi et al., 2009) intractable epilepsy, 50–80% of all seizure recurrence occurs within the first two to six postoperative months, followed by a much slower 3–5% subsequent yearly decrease in seizure-freedom rates. This variation in the rate of seizure recurrence likely reflects shifts in the mechanisms of recurrence. In fact, whenever the predictors of these two phases of recurrence (early vs. late) were analyzed independently for TLE surgery (Jeha et al., 2006), the strongest predictors of early recurrence were markers of an inaccurate localization of the epileptogenic zone or its incomplete resection such as residual spiking on postoperative EEG, bitemporal MRI abnormalities, and the need to use invasive EEG recordings. In contrast, late seizure recurrences were highest in patients whose pathology was restricted to nonspecific gliosis (Jeha et al., 2006). Furthermore, when compared with early surgical failures, those later recurrences manifested with less frequent seizures (Jeha et al., 2006, 2007) that were more readily controlled with medical therapy alone (McIntosh et al., 2005; Jehi et al., 2010a,b), behaving then more like a “new onset epilepsy” rather than a persistent focus of intractable seizures that was incompletely resected. This suggests that the mechanism of late recurrence may extend to include the maturational process of epileptogenesis required for seizures to develop again years after an initial period of postoperative seizure freedom. Within this framework, our current finding that a potentially antiepileptogenic medication may improve seizure outcomes, particularly late after surgery, is not surprising.