Use of pre-treatment MRI scans to predict toxicity from radiotherapy in patients with Glioblastoma (GB)

Abstract

Abstract Median life expectancy of patients with Glioblastoma (GB) is 24.4 weeks. Current best treatment involves concurrent use of Temozolomide and radiotherapy. However, toxicity from radiotherapy can limit patient benefit. The objective of this study was to determine the ability of pre-treatment MRI scans of the brain to predict radiotoxicity. This pilot study collected survival and toxicity data from 10 patients aged 65 years or older with a histopathological or radiological diagnosis of GB. Brain MRI scans were used to collect information on tumour characteristics alongside imaging scoring systems to assess medial temporal lobe atrophy, global cortical atrophy and small vessel disease. Scans were analysed by two independent readers. Median survival for this cohort was 35.2 weeks (range 19–59.3 weeks). The beta correlation scores for total contralateral white matter score, global cortical atrophy and T1 contrast volume and survival were all found to equal 1. Beta-correlation coefficients for microanatomical rating score and Schelten’s medial temporal atrophy score were 0.4 and 0.9 respectively. The sum of CTCAE scores for nausea, fatigue and confusion was calculated to provide an overall toxicity score. The mean beta-correlation co-efficient for all scores against toxicity was 1. Use of Bland-Altmann plots showed significant inter-rater reliability for all scores excluding global cortical atrophy. Results from this study suggest a moderate relationship between scan characteristics and increasing toxicity. However, high inter-rater reliability suggests these scoring systems have the potential to be used accurately in clinical practice. A larger study will be conducted to confirm results from this study.