This is the first study investigating the effect of invasive VNS and taVNS on verbal memory performance in drug-resistant epilepsy patients in a prospective setting. Immediate recall and delayed recognition scores were not significantly changed by acute VNS or taVNS. We did find an improvement in verbal memory performance after 6 weeks of VNS treatment irrespective of the acute intervention.
In contrast to our hypothesis, immediate recall and delayed recognition scores were not improved by acute invasive VNS and we were therefore unable to demonstrate an intensity-dependent effect of VNS as previously described [12]. No significant differences in attention scores were observed between the interventions, therefore excluding attentional deficits as a reason for these negative findings. Clark et al. demonstrated a significant effect of moderate amplitude (= 0.5 mA) VNS in drug-resistant epilepsy patients with a similar word recognition memory paradigm [12]. Different inclusion criteria were implemented in the study by Clark et. al. that may explain the contrasting findings. As opposed to the Clark study, many of our patients were treated with benzodiazepines and/or more than 2 anti-epileptic drugs. As polytherapy and benzodiazepines have been found to induce cognitive side-effects [2], this may have interfered with a potential beneficial effect of VNS. Clark et al. may also have included patients with relatively more severe epilepsy as patients with less than 4 seizures a month were excluded. In contrast, 6 patients in our study reported less than 4 seizures a month during baseline. Only patients with an IQ score below 70 were excluded in our study. Due to the aim of investigating both invasive VNS and taVNS in the first session of our study, the total duration of the experiment was longer (4 hours vs 2.5 hours in the Clark study) which may have led to a retention interval too long to reliably evaluate memory performance. We did indeed find a significant effect of order in the first session indicating that the most recently memorized words were more easily recognized. Only 2 experimental conditions were investigated in the second session reducing the total duration to 2.5 hours. In this shorter session, no recency effect was seen, demonstrating a more reliable evaluation of memory performance at shorter retention intervals. However, as the experimental conditions were counterbalanced across the experimental blocks and no effect of the interaction intervention*order was seen, the longer duration of session 1 alone does not explain our negative findings.
Secondly, we did not find a significant effect of taVNS on verbal memory performance. Moreover, a trend for lower delayed recognition scores was seen after taVNS compared to VNS ON (hit score) and VNS OFF (discrimination index). These findings are in line with the previous study by our group in healthy volunteers, demonstrating that acute taVNS, delivered for 30 s during the consolidation phase of a memory task, did not improve verbal memory performance [49]. We hypothesized the negative findings in the latter study were due to the study population of healthy volunteers who would be less prone to improvement by taVNS compared to epilepsy patients with a lower baseline performance. The current study argues against this hypothesis. Potentially, the duration of stimulation may have been too short to induce significant effects on verbal memory performance. In the study by Jacobs et al., an enhanced associative memory performance was seen after 17 min of continuous taVNS at the tragus [47]. However, as the tragus is not exclusively innervated by the auricular branch of the vagus nerve, this finding is not necessarily related to the activation of the vagal pathway [50]. Giraudier et al. demonstrated an improved recollection-based memory performance after taVNS was delivered in ON/OFF cycles of 30 s before, during and after an encoding task with a total duration of stimulation of 23 min [48]. These longer and/or continuous stimulation protocols may be more effective in modulating verbal memory performance than the short stimulation duration of 30 s implemented in our study. Additionally, only associative and recollection-based memory were affected in the study by Jacobs et al. and Giraudier et al. respectively, indicating that taVNS may affect specific processes of memory formation confined to specific pathways or brain structures instead of inducing a general cognitive improvement.
The taVNS stimulation current was set for each patient individually using the pain threshold method. Therefore, patients were stimulated with different output currents varying from 0.90 to 5.00 mA. As an intensity-dependent modulation of verbal memory performance has been described with VNS, we investigated whether taVNS stimulation current intensity was associated with verbal memory performance. No association was found between stimulation current and verbal memory performance scores. However, stimulation current was compared between subjects whereas a within subject analysis would be the most appropriate method to establish a dose-response relationship. The evaluation of different stimulation currents within one subject was however not feasible in the design of this study.
Thirdly, we demonstrated an improvement in both immediate recall and delayed recognition scores after 6 weeks of VNS treatment. This finding indicates that longer and more repetitive stimulation of the vagus nerve may be required to induce long-term potentiation and effectively modulate memory performance. Only a few studies have investigated the long-term effects of VNS on cognitive functioning in epilepsy patients. Although subjective cognitive improvements have been reported [7, 8], no objective changes in neuropsychological testing were found at 3 to 6 months follow-up [15–17]. This is the first study evaluating verbal memory performance after a short follow-up period during which low stimulation currents were used (0.5–0.75 mA). A longer follow-up would be needed to evaluate if this effect is transient or restricted to moderate stimulation currents, explaining the negative findings in the studies with longer follow-up periods. These results may indicate that the modulatory effect of VNS on memory performance occurs shortly after VNS therapy initiation and is more efficient at moderate stimulation intensities. Of note, due to the relatively short interval between both sessions, practice effects may also have contributed to the enhanced performance in the second session and should be taken into account in the interpretation of these findings.
No effect of clinical response on verbal memory performance was seen, indicating that VNS responders did not obtain higher verbal memory performance scores after 6 weeks of VNS compared to non-responders. This suggests that the memory enhancing and seizure suppressing effects of VNS engage different mechanisms and potentially require distinct stimulation protocols. The effect of long-term taVNS on memory performance has not been investigated to this date and should be further explored.
Finally, this study has some limitations. First, the 60-minute breaks between the experimental conditions may have been too short to ensure complete wash-out. Studies investigating the enduring effects of VNS on NE have shown inconsistent results ranging from completely transient effects [23] to elevated NE levels up to 2 h after stimulation [20]. To our knowledge, the enduring effects of taVNS have not been investigated. However, increasing the wash-out periods would have prolonged the duration of the experimental session inducing an even longer retention interval with potential order effects as well as other confounders such as fatigue, diurnal variability, hypoglycemic and post-prandial effects. A second limitation of this study is the incomplete blinding. Patients could easily distinguish the VNS OFF condition from the VNS ON condition, reflected by a significant difference in pain reports after stimulation. In addition, patients were not blinded for the taVNS condition as no sham taVNS condition was included. However, even with incomplete blinding, patients did not perform better after the active stimulation conditions. As a significantly higher pain score was reported after taVNS and VNS ON compared to VNS OFF, the experience of pain during stimulation could potentially decrease memory performance. However, as pain reports were generally low, it seems unlikely this would explain our negative findings. Finally, the finding of an enhanced verbal memory performance after 6 weeks of VNS was not controlled as all patients received active VNS. A randomized controlled trial would provide a higher level of evidence as this would control for placebo effects, practice effects and/or differential experimental circumstances.