Inter-observer variability in the measurement of diffuse intrinsic pontine gliomas

Abstract

Diffuse intrinsic pontine glioma (DIPG) is an invasive pediatric brainstem tumor with a poor prognosis. Patients commonly enter investigational trials, many of which use radiographic response as an endpoint for assessing drug efficacy. However, DIPGs are difficult to measure on magnetic resonance imaging (MRI). In this study, we characterized the reproducibility of these commonly performed measurements. Each of four readers measured 50 MRI scans from DIPG patients and inter-observer variability was estimated with descriptive statistics. Results confirmed that there is wide variability in DIPG tumor measurements between readers for all image types. Measurements on FLAIR imaging were most consistent. For patients on clinical trials, measurement of DIPG should be performed by a single reader while comparing prior images side-by-side. Endpoints for clinical trials determining efficacy in this population should also include more objective measures, such as survival, and additional endpoints need to be investigated.