P11.03 Sleep analysis of accelerometer data from High Grade Glioma patients in the BrainWear study

Abstract

Abstract BACKGROUND Sleep disturbance is a common symptom in patients with high grade glioma (HGG). Existing self-reported and uni-dimensional data from questionnaires are of limited value. The observational phase 2 trial, BrainWear (ISRCTN 34351424) provides the first objective analysis of sleep in HGG patients. MATERIAL AND METHODS Patients with HGG were asked to wear an AX3 Axivity tri-axial accelerometer throughout treatment. The study employed a wear-as-long-as-possible approach to accelerometry data collection, and we used age-sex matched controls from the UK Biobank as comparators. Baseline was established as a 7 day period of wear prior to surgery or at least 7 days post-surgery. The dataset for this analysis consists of 21 patients with data at baseline and 15 patients during chemo-radiation. Only 16 of the 21 HGG patients at baseline were included for initial comparisons with healthy controls due to age limitations of the UK Biobank cohort for matching. Raw accelerometer data was processed using the GGIR package, with non-imputation of missing data, exclusion of days with <16 hours of wear time and removal of algorithm-identified problematic data. Mann-Whitney U-tests and unpaired T-tests were used to compare 7 sleep-related features between HGG patients and healthy controls at baseline, with choice of statistical test based on shapiro-wilk derived normality. Secondly, to assess changes in sleep in HGG patients across treatment period, K-means clustering of 5 sleep parameters, available longitudinally, was conducted to explore sleep behaviours at baseline (n = 21) and during chemo-radiation. RESULTS HGG patients (n = 16) exhibited greater daytime inactivity than healthy controls (n = 32) (p < 0.0001, 2.2 vs 0.5 hrs) and more variation in their 24 hour activity rhythm from day to day (p < 0.0001, 0.12 vs 0.18). We identified 5 sleep features which allowed us to cluster patients’ sleep behaviour, and most (62.5%) of HGG patients have a poor sleep profile. This sleep profile was characterised by an average of 5.4 hours of night-time sleep, 2.1 hours of daytime inactivity and disturbed sleep quality. However, evaluation of HGG patient sleep cluster designation at baseline and during chemoradiation, showed HGG patients with data at both timepoints (n = 9) demonstrate stability or improvement in sleep profile. CONCLUSION Patients with HGG have objective evidence of poor sleep compared to healthy matched controls. Further work will explore changes in sleep over time.