NIMG-10. BRAIN TUMOR MRI STRUCTURED REPORTING ALLOWS CALCULATION OF INTERRATER AGREEMENT OF PATIENTS REVIEWED AT TUMOR BOARD. (11th November 2019)

• Record Type: Journal Article
• Authors: Cooper, Maxwell; Hoch, Michael; Abidi, Syed; Hu, Ranliang; Shim, Hyunsuk; Weinberg, Brent
• Is Part Of: Neuro-oncology. Volume 21(2019)Supplement 6
• Page Start: vi163
• Page End: vi163
• Publication Date: 2019-11-11
• Subjects: Brain Neoplasms -- Periodicals; Brain -- Tumors -- Periodicals; Brain -- Cancer -- Periodicals; Nervous system -- Cancer -- Periodicals; 616.99481
• DOI: 10.1093/neuonc/noz175.682
• Languages: English
• ISSNs: 1522-8517

Abstract

Abstract: INTRODUCTION: Longitudinal imaging evaluation of brain tumor patients with MRI is a challenging endeavor, as there is considerable overlap between imaging findings of tumor progression and treatment related changes. There is felt to be a high degree of variation between individual interpreters, particularly in complex cases. Structured reporting systems, such as the recently described brain tumor reporting and data system (BT-RADS), attempt to maximize consistency between reading radiologists, although how successful they are at achieving this is unknown. METHODS: Patients with a diagnosis of a primary brain tumor with imaging presented at adult brain tumor board between October 2017 and March 2019 were reviewed. The most recent follow-up MRI was scored on an 8-point scale as described by BT-RADS, ranging from 0 to 4 with increasing suspicion for worsening disease, as described in the original radiologist report. Secondary review of MRI was performed by one of three neuroradiologists routinely participating in brain tumor conference. Interrater agreement between primary and secondary review were calculated using rates of exact agreement and linear weighted kappa. RESULTS: Of studies reviewed, 270 out of 275 had imaging and reports suitable for repeat structured scoring. Grade 4 astrocytoma (glioblastoma) was the most common diagnosis, followed by grade 2 oligodendroglioma and grade 2 astrocytoma. The overall agreement rate between initial and secondary review was 83.0% with kappa of 0.89 +/- .05. There was perfect agreement on studies with improvement (1a or 1b) with lower levels of agreement (63.6–82.4%) for studies with worsening imaging findings (3a-4). CONCLUSIONS: Using a structured reporting system to categorize MRIs in brain tumor patients allows for precise characterization of variation between report conclusions on primary and secondary review. Overall agreement is good, but variation rates increase with worsening findings which may be harder to correctly interpret.