BrainWear: Longitudinal, objective assessment of physical activity in 42 High Grade Glioma (HGG) patients. (15th October 2021)

• Record Type: Journal Article
• Authors: Dadhania, Seema; Pakzad-Shahabi, Lillie; Le Calvez, Kerlann; Saleem, Waqar; Wang, James; Mohammed, Waleed; Mistry, Sanjay; Williams, Matthew
• Is Part Of: Neuro-oncology. Volume 23: Supplement 4(2021)
• Page Start: iv3
• Page End: iv3
• Publication Date: 2021-10-15
• Subjects: Brain Neoplasms -- Periodicals; Brain -- Tumors -- Periodicals; Brain -- Cancer -- Periodicals; Nervous system -- Cancer -- Periodicals; 616.99481
• DOI: 10.1093/neuonc/noab195.006
• Languages: English
• ISSNs: 1522-8517

Abstract

Abstract: Aims: In patients with HGG, we know that QoL and physical function decline with progressive disease (PD) and fatigue is a strong predictor of survival in recurrent disease. Despite notable technical advances in therapy for in the past decade, survival has not improved. The role of physical function as a predictor of QoL, treatment tolerance and as an early indicator of worsening morbidity (e.g. tumour recurrence) is an area of growing importance. Recent advancements in wearable technology allow us the opportunity to gather high-quality, continuous and objective data BrainWear is a feasibility study collecting longitudinal physical activity (PA) data from patients with primary and secondary brain tumours and we hypothesise changes in PA over time, are a potentially sensitive biomarker for PD both at diagnosis and relapse. Method: Here we show early analysis of this novel dataset of 42 HGG patients and will present: 1) feasibility and acceptability 2) how digitally captured PA changes through treatment and at PD/hospitalization 3) the correlation between patient reported outcomes (PRO) and PA data 4) how PA in HGG patients compares with healthy UK Biobank participants. PA data is collected via a wrist-worn accelerometer. Raw accelerometer data is processed using the UK Biobank Accelerometer Analysis pipeline in python 3.7, and evaluated for good quality wear-time. Overall activity is represented as vector magnitude in milligravity units(mg) and a machine-learning classifier classifies daily activity into 5 separate groups (walking, tasks-light, moderate, sedentary and sleep). Descriptive statistics summarise baseline characteristics and unadjusted mean used to present vector magnitude and accelerometer-predicted functional behaviours (in h/day) by age, sex, radiotherapy and weekend days. Mixed effect models for repeated measures are used for longitudinal data evaluation of PA. Results: Between October 2018 and March 2021, 42 patients with a suspected HGG were recruited; 16 females and 26 males with a median age of 59. 40 patients had surgery and 35 patients had adjuvant primary radiotherapy, 23 of whom had a 6-week course. They have provided 3458 days of accelerometer data, 80% of which has been classified as good quality wear-time. There are no statistical differences in mean activity between gender, patients >60 years show statistical difference in time spent doing moderate activity compared to those <60 years, and there are significant differences in mean vector magnitude and walking between radiotherapy and non-radiotherapy days. In patients having a 6-week RT course, time spent in daily moderate activity falls 4-fold between week 1 and the second week following RT completion (70 minutes to 16 minutes). HGG versus healthy UK Biobank participants shows significant differences in all measures of PA. Conclusion: Here we present preliminary analysis of this highly novel dataset in adult high grade glioma patients, and show digital remote health monitoring is feasible and acceptable with 80% of data classified as high quality wear-time suggesting good patient adherence. We are able to objectively describe how PA changes through standard treatments and understand the inter and intra-patient variation in PA, and whether there are correlates with patient-centred measures, clinical measures and early indicators of worsening disease. We will present further data on changes in PA prior to hospitalisation and at disease progression, and discuss some of the challenges of running a digital health trial. The passive and objective nature of wearable activity monitors gives clinicians the opportunity to evaluate and monitor the patient in motion, rather than the episodic snapshot we currently see, and in turn has the potential to improve our clinical decision making and potentially outcomes.