Flat Out Unnecessary: Burst Characteristics, Not Duration of Interburst Intervals, Predict Successful Anesthetic Wean in Refractory Status Epilepticus

Commentary

EEG Characteristics of Successful Burst Suppression for Refractory Status Epilepticus.

Johnson EL, Martinez ND, Ritzi EK. Neurocrit Care 2016;25:407–414.

BACKGROUND: Refractory status epilepticus (RSE) is often treated with continuous intravenous medications with the goal of EEG burst suppression. Standard advice is to titrate medications to at least 10-s interburst intervals; however, this has not been shown to improve outcome. We examined EEG characteristics in patients treated with IV anesthetic therapy (IVAT) for RSE to determine which EEG characteristics were associated with successful lifting of IVAT (i.e., without recurrence of status epilepticus). METHODS: We screened the clinical continuous EEG database for adult patients treated with IVAT for RSE (excluding patients with anoxic injury). We measured the length of bursts and interburst intervals for each patient, calculated EEG burst suppression ratios, and graded bursts for the amount of epileptiform activity. We compared these characteristics in successful versus unsuccessful IVAT lifting attempts. RESULTS: We included 17 successful and 20 unsuccessful lifting attempts in 19 patients (5 used as a holdout validation set). The interburst intervals, burst suppression ratios, and length of bursts did not differentiate successful and unsuccessful lifting attempts; the amount of epileptiform activity in bursts correlated with success or failure to wean IVAT (p = 0.008). Maximum burst amplitude <125 μV had 84.6 % sensitivity and 61.1 % specificity for predicting successful lifting. CONCLUSION: The length of interburst intervals and burst suppression did not predict successful termination of RSE in this small cohort. This may suggest that EEG characteristics, rather a strict interburst interval goal, could guide IVAT for RSE.

Refractory status epilepticus (RSE), defined as failure to respond to first- and second-line treatments, carries considerable risks, with in-hospital mortality surpassing 60% (1). Despite the severity of the condition, the only class I evidence to guide treatment of status epilepticus is the recommendation to use benzodiazepines as first line (2). Typically, antiepileptics, such as fosphenytoin, levetiracetam, and valproic acid follow as second line, yet there is inadequate evidence to guide best practice (3). Intravenous anesthetic therapy is often employed as treatment for RSE with the standard recommendation to achieve burst suppression with interburst intervals of 10 seconds for at least 24 hours before weaning therapy (4). We've largely accepted the criteria for adequate burst suppression on faith, as there is essentially no evidence to guide how much suppression is truly needed to achieve favorable outcomes. We know that outcome following RSE is highly related to the underlying etiology, so, to some degree, the ability to improve outcome with burst suppression may be out of our hands (5). Use of anesthetic-induced burst suppression comes with considerable potential adverse effects, which could contribute to morbidity; therefore, any markers that guide how aggressive burst suppression should be pursued will likely improve outcomes in the long run.

Johnson et al. evaluated the continuous EEGs of 19 patients with RSE to investigate whether longer durations of interburst intervals increased the likelihood of weaning from anesthetic therapy and whether EEG characteristics of the bursts could predict successful weans. The authors analyzed 10 consecutive bursts and suppression intervals in a representative sample of the EEG 12 hours prior to anesthetic wean, thus a method that could easily be applied in daily practice. They found no benefit to deeper levels of burst suppression (i.e., interburst intervals of 10 seconds or longer), but showed that bursts without monomorphic sharp waves, with amplitudes less than 125 μV, and containing less than 50% epileptiform activity predicted more successful weaning. They further validated their hypothesis by applying these criteria to five additional patients in whom they accurately predicted successful wean in 80% of the attempts. While the cohort was small, the ability to potentially utilize EEG characteristics to accurately predict successful weaning may allow anesthetic burst suppression to be tailored for the patient, instead of simply accepting a dogmatic definition of necessary burst suppression.

This is not to say that less is more for all patients. Studies of EEG-guided burst suppression as a predictor of outcome tend to be retrospective and rarely have included the same patient populations or variables, making comparisons difficult. In a prior study by Krishnamurthy and Drislane (6), patients titrated to “flat” suppression were more often seizure free and more likely to survive than those titrated to burst suppression. However, the same study contained three patients in RSE that were neither burst suppressed or “flat” (termed “slow”) and all three had seizure-free outcomes. This study did not specifically examine the EEG characteristics of the bursts and medication titration was at the discretion of treating physician (as in other studies). We have to wonder whether the three patients that were not fully suppressed had a similar favorable EEG pattern as seen in the study by Johnson, et al. The present study identified seven patients with RSE that were not burst suppressed and thus did not meet inclusion criteria. We are left to wonder whether these patients had similar favorable EEG characteristics that led to less aggressive treatment and potentially better outcome.

If we can identify EEG characteristics that predict weaning failure, might longer durations of treatment be more efficacious for selected patients when likelihood to fail is recognized early? Patients that failed initial wean in this study tended to have longer subsequent periods of suppression and trended to more favorable ability to wean in subsequent attempts, though the relationship to ultimate outcome was not examined. Some prior investigations have not found a connection between duration of burst suppression and outcome, while others suggest a negative correlation (6, 7). Until a study is designed using a treatment protocol based on EEG characteristics, this question will remain unanswered.

Many other questions remain in the treatment of RSE. While this study suggests EEG features that may guide how aggressively we pursue burst suppression, it does not show that limiting aggressive burst suppression will impact outcome. We still don't know if patients with unfavorable EEG characteristics may benefit from aggressive burst suppression even to the point of a “flat” EEG in order to “cool off” the brain. No study has standardized the anesthetic used, so we have no idea how different medications may impact the EEG, the likelihood of successful weaning, or the ultimate outcome. While etiology of RSE is a key predictor of outcome, we don't yet know if using a treatment protocol based on EEG characteristics and etiology might result in decreased morbidity.

This study represents yet another step in defining the variables essential to developing a prospective study of RSE treatment. Moving forward, we will benefit from studies that standardize the treatment of RSE based first on the characteristics of the EEG to guide a protocol-driven titration of anesthetic therapy with attention to ultimate outcome. Studies will need to be large enough to include a variety of etiologies in order to assess the contribution of cause to ultimate outcome. These results can then lead to refinements of the titration protocol based on etiology with more aggressive burst suppression reserved for those etiologies where the benefits will likely outweigh the risks of more aggressive therapy.