Repeatability and reproducibility of relative cerebral blood volume measurement of recurrent glioma in a multicentre trial setting.

Measurement of relative cerebral blood volume (rCBV) with dynamic susceptibility contrast (DSC) magnetic resonance imaging (MRI) is used extensively for brain tumour diagnosis and follow-up. The aim of this pilot study was to assess the robustness of rCBV measurement in patients with enhancing recurrent glioma in a European multicentre trial setting. We included pre-treatment postcontrast T1 weighted (T1w) and DSC scans of 20 patients with recurrent glioma from 2 European Organisation for Research and Treatment of Cancer trials (26101 and 26091). Three reviewers independently placed a fixed circular region of interest of 70 mm2 in the tumour area of highest rCBV (rCBV max). To calculate the normalised rCBV max (nrCBV max), three ROIs were placed in the anterior, middle and posterior centrum semiovale normal-appearing white matter of the contralateral hemisphere. After several months, each observer repeated the assessments blinded for initial findings. Repeatability and reproducibility were estimated with a mixed model. Each measurement was also classified according to 4 clinically meaningful categories. Three patients were post hoc excluded from analysis because of lack of enhancing tumour. The mean nrCBV max repeatability was 49.5%, and reproducibility was 5.5%. In 14 of 17 patients, at least 2 reviewers classified the patient into the same category. Our results indicate that a well-established review process needs to be applied upfront to assess perfusion in a multicentre trial setting. While awaiting further validation, we propose as a strategy to measure rCBV in the context of recurrent glioma trials to use two central reviewers and an adjudicator in case of disagreement. • Glioma perfusion measurement is only modestly repeatable and reproducible. • Most tumours were classified in the same perfusion category by 2 reviewers. • Relative cerebral blood volume reference values should be used with great care. • In clinical trials, categorisation should be performed by at least two readers. [ABSTRACT FROM AUTHOR]