The Best Things Happen When You Least Expect Them: Responsive Neurostimulation During Low-Risk Brain States is Associated with Improved Long-Term Seizure Suppression

Abstract

Closed-Loop Stimulation in Periods With Less Epileptiform Activity Drives Improved Epilepsy Outcomes

Anderson DN, Charlebois CM, Smith EH, Davis TS, Peters AY, Newman BJ, Arain AM, Wilcox KS, Butson CR, Rolston JD. Brain. 2024;147(2):521-531. doi:10.1093/brain/awad343. PMID: 37796038; PMCID: PMC10834245

In patients with drug-resistant epilepsy, electrical stimulation of the brain in response to epileptiform activity can make seizures less frequent and debilitating. This therapy, known as closed-loop responsive neurostimulation (RNS), aims to directly halt seizure activity via targeted stimulation of a burgeoning seizure. Rather than immediately stopping seizures as they start, many RNS implants produce slower, long-lasting changes in brain dynamics that better predict clinical outcomes. Here we hypothesize that stimulation during brain states with less epileptiform activity drives long-term changes that restore healthy brain networks. To test this, we quantified stimulation episodes during low- and high-risk brain states, that is, stimulation during periods with a lower or higher risk of generating epileptiform activity in a cohort of 40 patients treated with RNS. More frequent stimulation in tonic low-risk states, and out of rhythmic high-risk states, predicted seizure reduction. Additionally, stimulation events were more likely to be phase-locked to prolonged episodes of abnormal activity for intermediate and poor responders when compared to super responders, consistent with the hypothesis that improved outcomes are driven by stimulation during low-risk states. These results support the hypothesis that stimulation during low-risk periods might underlie the mechanisms of RNS, suggesting a relationship between temporal patterns of neuromodulation and plasticity that facilitates long-term seizure reduction.

Commentary

Responsive neurostimulation (RNS) has emerged as a valuable therapeutic tool for patients with epilepsy, particularly for the 30% of patients who do not respond to antiepileptic drugs.^1,2 RNS involves implanting a closed-loop neurostimulation device into the brain that detects epileptic activity in real-time and delivers immediate electrical current to suppress seizures. Overall, clinical outcomes from RNS have been very successful, with a median reduction in seizure frequency of 75%. ³ However, patient outcomes are highly variable, and only 18% of patients achieve at least 1 year of seizure freedom. ³ Furthermore, despite the overall effectiveness of RNS treatment in reducing seizure frequency, the specific mechanisms by which this effect is achieved remain unknown, as well as the factors that control whether or not a patient will respond to the treatment. To address this gap, Anderson et al ⁴ conducted a retrospective study on RNS patients in order to identify biomarkers that correlate with improved patient outcomes. The authors found that the effectiveness of RNS intervention was correlated with the proportion of stimulations that were delivered during interictal periods with less abnormal epileptiform activity. This work provides new evidence that RNS may not work by simply interrupting ongoing seizures, but rather provides long-term seizure suppression by creating new plasticity from the stimulations that occur during low-risk brain states.

The RNS device was designed to detect epileptiform activity and deliver stimulation in real-time with the goal of aborting spontaneous seizures. ⁵ However, there is now substantial evidence that the therapeutic mechanisms of RNS are unlikely to be through real-time seizure disruption. In particular, the therapeutic benefits of RNS take several months to manifest, the number of stimulations far exceeds the number of seizures per day, and a large proportion of stimulations are delivered during interictal periods.^4,6 Furthermore, clinical outcomes are correlated with long-term changes in the coordination of neural activity, ⁶ suggesting that new plasticity that reconfigures functional connectivity across the brain may be key to the therapeutic effects of RNS. Critically, the plasticity that underlies these changes in functional connectivity is more likely to be created during nonepileptic brain states, suggesting that electrical stimulations during a period of normal brain activity may actually be more therapeutic than stimulations during seizures.

Anderson et al ⁴ directly explored this hypothesis that RNS stimulations may be more effective during nonepileptic brain states by examining the relationship between patient outcomes and when stimulations occurred. The authors first separated patients into super responders (with >90% reduction in seizures), intermediate responders (seizure reduction between 50% and 90%), and poor responders (<50% reduction in seizures). In order to examine the relationship between seizure reduction and the epileptic state during which stimulations were delivered, the authors categorized stimulations as occurring during periods of either high-risk or low-risk of generating seizures, based on the amount of abnormal epileptic activity at each time point. They found that the proportion of stimulations delivered in low-risk states and the proportion of overall time spent in low-risk states were both highly correlated with a greater reduction in seizure frequency. Furthermore, they also found that patients with more consistent circadian cycling of epileptic activity had better long-term seizure suppression than patients with less consistent rhythms.

These findings have several important implications for the treatment of patients with RNS. First, the results support the authors' main hypothesis that electrical stimulations during low-risk, relatively normal brain activity may be more therapeutic than stimulations during high-risk periods with more epileptic activity. Notably, this hypothesis is entirely contrary to the conceptualization of RNS as a responsive stimulation system and suggests that the RNS device may be even more effective if the stimulations are delivered specifically during nonepileptic periods. This notion is further supported by deep brain stimulation studies, where stimulations are given without respect to brain state and have mostly comparable results to RNS patients. ⁷ Second, these results provide important clinical insight into the predictability of success with RNS treatment. Currently, there are no established methods to determine whether a patient will respond to RNS. Anderson et al ⁴ found clear evidence that the success of long-term seizure suppression can be predicted based on the amount of time spent in low-risk brain states and the consistency of circadian cycling. Critically, this prediction can be made well before the clinical benefits are observed and could be used clinically to select which patients are more likely to benefit from RNS therapy.

A major limitation of this work is the retrospective and correlational nature of the study. As the authors acknowledge, this work does not provide any causal evidence of the relationship between the brain state at which RNS stimulations occur and the effectiveness of the treatment. It is possible, then, that other related factors (such as the amount of underlying damage to the brain or the ability to produce plasticity) may be the critical factor in determining the success of long-term seizure suppression. In particular, the design of this study does not allow the authors to dissociate the impact of timing stimulations on low-risk states from the amount of time a particular patient spends in a low-risk state. While the authors found no relationship between baseline seizure frequency and RNS outcomes, patients who spend more time in low-risk states may still have less severe cases of epilepsy that are easier to treat with RNS. To directly assess the causal role of stimulation timing, it is necessary to perform new clinical trials with stimulations targeted to low-risk periods. With the now extensive evidence that long-term seizure suppression is linked to neural plasticity, these trials should be a high priority as they have a substantial potential impact on clinical outcomes for RNS patients.

This work also highlights the clear need for more mechanistic studies on seizure suppression with electrical stimulation. While RNS has been Food and Drug Administration approved for over a decade, there remains very little known about the specific mechanisms that drive its effects. There is now a wealth of data from RNS patients, but the limitations of retrospective human studies will always constrain the conclusions that can be drawn. Therefore, it is imperative to develop new animal models of RNS treatment where direct causal mechanisms can be studied and then applied to patients. Using animal models, it would be easy to test how the timing of stimulations impacts seizure suppression. Furthermore, there may be other brain states (such as at specific phases of theta oscillations) that provide windows of increased plasticity, and stimulations targeted to these periods may further increase the effectiveness of stimulation. Using animal models would also allow for the development of new open-source algorithms that can provide more transparent and sophisticated detection and stimulation patterns, which have a high potential to improve upon current recording and stimulation protocols.

Overall, this study from Anderson et al ⁴ provides valuable insights that can help guide future treatment of drug-resistant epilepsies. This work identified key biomarkers of RNS outcomes (time spent in low-risk states and the strength of circadian cycling) that can be used clinically to weigh the likelihood of successful outcomes with RNS against alternative treatment options, such as surgical resection. Furthermore, this work suggests that targeting RNS stimulation parameters to preferentially target interictal states may increase the effectiveness of RNS treatment. While this hypothesis still needs to be further tested, it has the potential to greatly improve outcomes for a large number of RNS patients with uncontrolled seizures.