This study provides insights into the global and local disturbances of spindles, the influence of sleep homeostasis, and the functional correlates of such disturbances in patients with various types of focal epilepsies. Using combined polysomnography and HD-EEG, we investigated global and local changes in rates of spindles in relation to the anatomical location of the epileptic focus and correlated these findings with neurocognitive performance. The most relevant findings of the current study are that (i) there is a reduction in spindle rates in epilepsy patients compared to healthy controls and a local decrease in spindle rates in the region with the epileptic focus compared to the contralateral side, (ii), there is a sleep homeostatic effect on spindle and spike rates across the night, and (iii) there is a negative correlation between a local reduction in spindles and neurocognitive performance measures.
Overall spindle rate and distribution are impaired in focal epilepsy
Our work is the first to show a reduction of spindles in patients with various types of drug-resistant focal epilepsy on a global and regional level compared to healthy controls and additionally at the site of the epileptic focus as compared to the contralateral homologous brain region. There are few studies in various epilepsy types using different methodology with only two studies having a healthy control group9,10,11,12,29. The evidence from these studies suggests that epilepsy impacts the presence of spindles, with most knowledge obtained in idiopathic generalized epilepsy and specific childhood epilepsy syndromes. A global reduction in spindle rates has been previously reported in a pediatric epilepsy population relative to a healthy control group10. Our work revealed a global reduction in spindle rates for adult patients with focal epilepsy compared to healthy controls over all N2 epochs with significant spindle reductions in both early and late N2 sleep. Although sleep macrostructure may have an influence on spindle distributions, this did not differ between epilepsy patients and controls. An exception to this was in the case of sleep latency, the amount of REM sleep in comparison to our healthy control group and the amount of N1, N3 and REM sleep in comparison to a larger healthy control population28. However, sleep latency was not reduced in epilepsy patients in comparison to the larger healthy control group of Mitterling et al.28 and the younger age of our study population in comparison to Mitterling et al.28 may explain the higher amount of N3 sleep. Overall, the spindle decrease is unlikely to be confounded by parameters of sleep macrostructure. Furthermore, it would not explain the focal reduction in spindles which was found at the site of the epileptic focus.
Within epilepsy patients, spindle rates were reduced in the region containing the epileptic focus. These findings, despite being generated in our heterogeneous group of focal epilepsy patients on variable medications, add new knowledge to the previous findings in children with childhood epilepsy with centrotemporal spikes12. Further, the local reduction in spindle rates was particularly pronounced for fast spindles. Interestingly, fast spindles play an important role in cognition7 and are reduced in other neurocognitive diseases such as Alzheimer’s disease30 and schizophrenia8. Although exploratory, our findings support growing evidence suggesting that local brain pathology, such as that which may be present in epilepsy, is not only associated with an increase in pathological markers of disease such as epileptic spikes or pathological slow waves, but also with a reduction of physiological activity such as spindles. This latter hypothesis is in keeping with an intracranial EEG study which showed that there is an inverse relationship between epileptic spikes and spindles in the ipsilateral hippocampi of patients with temporal lobe epilepsy31.
Fast spindle reduction is correlated with cognitive performance
Local reductions in fast spindles were correlated with lower cognitive performance such as speed performance. This is the first study in a group of heterogenous focal epilepsy patients showing a correlation between spindle decreases and neurocognitive function. Our results are consistent with a previously demonstrated relationship between fast spindle rates and neuropsychological performance7,8 and with findings of negative correlations between spindle rates in the area of the epileptic focus and intelligence quotient and executive functions in childhood epilepsy with centrotemporal spikes12. In the current study, focal epilepsy patients with more pronounced spindle reductions showed greater decreases in speed performance, whereas there was no effect on accuracy. Patients showed this reduction in both subscales measuring speed (controlled search and automatic detection). This points to the fact that the spindle decrease might have a more general effect on psychomotor processing speed. Spindles have their maximum over the fronto-centro-parietal regions32 and around 50% of spindle events are symmetrically distributed over both hemispheres23. Given that fast spindles are associated with processing speed8, a reduction of spindles in the brain region with the epileptic focus may therefore have an impact on processing speed due to the imbalance of spindle distribution. On the verbal memory task, only the subgroup of patients with left temporal lobe epilepsy showed a negative correlation between spindle reductions and memory performance. The relationship between impairment of verbal memory in temporal lobe epilepsies is well known33. Nevertheless, the concurrent spindle reductions and decreased neuropsychological parameters found in our study does not allow the presumption of a causal link between those two factors.
Temporal and extratemporal lobe epilepsies have different spindle reduction profiles
Patients with extratemporal lobe epilepsy had significant decreases in all spindle and fast spindle rates in the region with the epileptic focus. Fronto-centro-parietal brain areas are known to have maximal rates of spindles compared to other brain regions32, thus it is possible that electrophysiological investigations in extratemporal epilepsies are better able to capture reductions in spindle rates. In contrast, patients with temporal lobe epilepsy showed only a decrease in fast spindle rates in the region with the epileptic focus.
Sleep homeostatic properties of spindles and spike rates
It is well known that homeostatic processes are implicated in sleep regulation34. In our study, there was a homeostatic spindle profile, with more regional reduction of spindles in early compared to late N2 sleep. Interestingly, the decrease in fast spindles in patients with epilepsy was sustained throughout the night. Due to the correlation of fast spindles with cognitive performance7, this reduction in spindles may underlie, at least in part, the cognitive comorbidities frequently encountered in patients with epilepsy35.
Spike rates were higher in the beginning of the night compared to the end of the night in patients with epilepsy. This points to the fact that spike rates are under the control of sleep homeostasis processes. In children with electrical status epilepticus in sleep, the overnight decrease in sleep slow waves is higher outside the epileptic focus compared to the decrease in the area of the epileptic focus36. Therefore, epileptic activity seems to influence the homeostasis of physiological sleep oscillations such as sleep slow waves depending on the location of the epileptic focus. Conversely, spikes are modulated by sleep slow waves, and we recently showed that spikes and high-frequency oscillations, used as novel markers of the epileptogenic zone, were strongly correlated with higher amplitude slow waves4. Given that slow wave power is higher in early compared to late sleep37, a higher spike rate in early sleep can be well explained. To the best of our knowledge, the only evidence pointing to an overnight decrease in spikes is from children with epilepsy showing lower centrotemporal spike rates towards the end of the night38. A potential explanation for the more pronounced decrease in regional spindles in early versus late N2 sleep could be a higher rate of spikes in the beginning of the night. This was, however, not confirmed in the present study, as we found no correlation between the spike rate and the regional reduction in spindles. This is potentially explained by the underlying neuropathology in our sample of patients with epilepsy39 which may have caused the disruption in spindle generation and cognitive function.
Reduction of sleep physiology as a potential marker of the epileptogenic zone
Until now, biomarkers used to define epileptogenicity are based on pathology such as the presence of epileptic spikes, high-frequency oscillations as novel biomarkers of the epileptogenic zone, or pathological slow waves40. Our work calls for a paradigm shift whereby clinicians should consider not only the presence of pathological markers but the absence of physiological markers to aid in the identification of the seizure focus in patients with epilepsy. Based on our findings, the local reduction of sleep spindles could be further explored not only as a marker of the epileptic focus but also as a follow-up marker after medical or surgical intervention, and postsurgical persistence of local reductions may be predictive of seizure relapse.
One main limitation of the study is that our study does not confirm whether spindles or an underlying third factor are the primary driver of cognitive dysfunction in patients with epilepsy. As known from intracranial EEG studies, not all spikes, particularly those from deep foci, can be recorded from the scalp15. Furthermore, while spindle rate is a stable intra-individual characteristic, variations in spike rate can occur from night to night41. This might also potentially be a reason for not finding a correlation between spike and spindle rates. Additionally, spindle analyses were performed solely in N2 sleep. This was decided based on the fact that spindles are more frequent in N2 sleep than in N3 sleep23. Further, the results were not corrected for multiple comparisons due to the exploratory nature of our study. The number of participants with available data for attention performance was low with no possibility to compare the neuropsychological performance to a healthy control group. Our findings shall be therefore strengthened in future studies with a larger sample size and neuropsychological testing in a control group. Finally, a previous study found that spindle rates were 0.16 times higher in women23. There was a higher percentage of male subjects in our control group (75%) compared to the epilepsy group (43%), so it is possible that differences in spindle rates between these groups may have been underestimated due to the gender characteristics of our study’s samples.