The prognostic value of sleep spindles in long-term outcome of West Syndrome

Commentary on

Spenner B, Krois-Neudenberger J, Kurlemann G, Althaus J, Schwartz O, Fiedler B. The prognostic value of sleep spindles in long-term outcome of West Syndrome. Eur J Paediatr Neurol. 2019 Nov;23(6):827-831. doi: 10.1016/j.ejpn.2019.09.003. Epub 2019 Sep 7.

An-Sofie Schoonjans and Karen Skjei

West Syndrome (WS) is a devastating infantile-onset epilepsy syndrome characterized by the triad of epileptic spasms, hypsarrhythmia on EEG and developmental delay or regression. As a prototypical epileptic encephalopathy, i.e. a condition in which "the epileptiform EEG abnormalities themselves are believed to contribute to a progressive disturbance in cerebral function” [1], resolution of hypsarrhythmia would be hypothesized to improve neurodevelopmental outcome.  Thus, a normalization of the EEG, whether defined as the suppression of abnormal patterns or the appearance of normal EEG patterns, could serve as a prognostic biomarker in WS.

In 1997, Kramer et al. investigated this hypothesis, assessing sleep architecture in the pretreatment EEGs of 53 WS patients.[2] They reported absent normal sleep patterns (sleep spindles or unequivocal K complexes) in 69%, however this did not correlate with developmental or seizure outcomes. In contrast, Lee et al. reported that among 66 WS patients, those who were seizure free for at least one year prior to the last follow-up visit were more likely to have had sleep spindles at presentation.[3] However, this finding was confounded by the fact that the seizure-free group had a significantly shorter lag time between onset of spasms and initiation of treatment, potentially affecting both the likelihood of spindles being present and patient outcome.

In the current paper, the authors attempted to clarify the relationship between sleep background and long-term outcome in WS by examining 448 serial sleep EEGs in 44 WS patients with 24-34 years’ follow-up. They hypothesized that an early resolution of hypsarrhythmia and the early recurrence of sleep spindles would be associated with a favorable long-term prognosis. The authors categorized outcomes using the Functional Independence Measurement (FIM) score as either independent (good outcome), partially dependent (intermediate outcome) or fully dependent (poor outcome) at last follow-up.

Among the 25 patients with absent sleep spindles on first EEG who had subsequent return of sleep spindles, the authors found no correlation between time to development of first sleep spindle and ultimate outcome category. They did not comment on outcome comparisons between patients with vs. without sleep spindles at presentation, nor between those without sleep spindles on first EEG who either did or did not regain their spindles.

In a subgroup analysis of the 13 patients with cryptogenic etiology (defined as those patients with no clear cause for their WS), they found that development of sleep spindles at any time was associated with outcome in Groups A or B (i.e. independent or partially dependent).

To assess the relationship between resolution of hypsarrhythmia and outcome, the authors divided time to cessation of hypsarrhythmia into 15 day intervals after presentation. They then performed descriptive statistics for each interval across the outcome groups.  The only statistically significant correlation with long-term outcome was found for cessation of hypsarrhythmia with 30 days after onset, which appeared to be more likely to be associated with good or poor (but not intermediate) outcome.

From these findings the authors concluded that neither recurrence of sleep spindles nor cessation of hypsarrhythmia were valid biomarkers of neurodevelopmental outcome in non-cryptogenic West Syndrome.

The study had several limitations. While hundreds of EEGs were reviewed, the overall number of patients was small, a common problem in studies in rare epilepsies. ‘Age at onset’ is defined as the first appearance of hypsarrhythmia on a sleep or wake EEG, but the times between first spasm and first EEG and between first spasm and treatment (a confounding factor in the Lee study, known to influence outcome)[4] were not mentioned nor included in the analyses. The inclusion/exclusion criteria used for an interpretation for hypsarrhythmia were not stated. And several interesting analyses (as mentioned previously) were either not performed or not commented on. The use of FIM as the primary outcome measure, while pragmatic, did not take into account seizure outcome, and its cognitive assessment is limited. And while the duration of follow-up was formidable, there was a large attrition of patients, with 38 patients (42% of their original population) either deceased (31 patients) or otherwise lost to follow-up (7 patients). Their EEGs should have been included in analysis, perhaps as a fourth outcome category.

A primary concern is the way the data were analyzed. While we are not statisticians, it seems that several continuous variables (e.g. FIM score, time to recurrence of spindles, time to resolution of hypsarrythmia) were divided into apparently non-ranked categories (e.g. outcome group, time intervals) and then dozens of comparisons made without correction for multiple analyses. While this may have been performed due to the small dataset, it can lead to false positives. We would love to get our hands on that dataset and sit down with a statistician to see what we could do!

To conclude, there is a need for prognostic biomarkers in West Syndrome. In this study they attempted to correlate the time to reappearance of sleep spindles among 25 WS patients as a prognostic factor for the outcome. Unfortunately, due to several limitations it is difficult to draw any conclusion from the data as analyzed. Larger, prospective studies with standardized treatments and follow-up would likely provide greater insight, both for West Syndrome and perhaps epileptic encephalopathies as a whole. 

References

1. Scheffer, I.E., et al., ILAE classification of the epilepsies: Position paper of the ILAE Commission for Classification and Terminology. Epilepsia, 2017. 58(4): p. 512-521.
2. Kramer, U., W.C. Sue, and M.A. Mikati, Hypsarrhythmia: frequency of variant patterns and correlation with etiology and outcome. Neurology, 1997. 48(1): p. 197-203.
3. Lee, Y.J., et al., Relationship between initial electroencephalographic characteristics and seizure outcomes in children with non-lesional West syndrome. Epilepsy Res, 2015. 110: p. 49-54.
4. O'Callaghan, F.J., et al., Safety and effectiveness of hormonal treatment versus hormonal treatment with vigabatrin for infantile spasms (ICISS): a randomised, multicentre, open-label trial. Lancet Neurol, 2017. 16(1): p. 33-42.

Article Abstract

Objective: There is a high risk for a profound developmental disorder in West Syndrome. However, a prognostic biomarker for neurodevelopmental outcome does not exist. Hypsarrhythmia disturbs normal EEG sleep patterns and hence sleep spindles, which are thought to be important for memory consolidation and learning. We postulated that the early recurrence of sleep spindles as well as an early resolution of hypsarrhythmic patterns after onset of West Syndrome lead to a favourable long-term outcome.

Method: 448 sleep EEGs recorded during the first two years of life in 44 patients with newly diagnosed West Syndrome between 1980 and 1989 were reviewed retrospectively. Long-term outcome was assessed in 2015–2016 by the Functional Independence Measurement Score as an indicator for coping with everyday situations. EEG-data were correlated with long-term outcome by Fisher's Exact Probability Test or Kruskal–Wallis H test.

Results: There were no statistically noticeable differences between time to cessation of hypsarrhythmia and long-term outcome. In a subgroup analysis of patients with cryptogenic etiology only (n = 13) recurrence of sleep spindles correlated with better long-term outcome (p = 0.038). In this subgroup 11/13 showed recurrence of sleep spindles in childhood, 10 of which had a good or intermediate outcome. Considering the whole patient cohort, recurrence of sleep spindles showed a statistically non-significant better long-term outcome.

Conclusion: Recurrence of sleep spindles and cessation of hypsarrhythmia cannot be used as a valid prognostic biomarker of neurodevelopmental outcome in non-cryptogenic West Syndrome.