Prognostic Value of Conventional Magnetic Resonance Imaging for Adult Patients with Brain Tumors

Abstract

Purpose: Magnetic resonance imaging (MRI) is the pivotal diagnostic step in patients with brain tumors, and is performed before histological diagnosis is available. We hypothesized that conventional MRI is as accurate as tumor histology in differentiating malignant from benign clinical course.

Methods: Two neuroradiologists blinded to any clinical information evaluated the first diagnostic MRI of 244 brain tumor patients before any treatment, using a self-developed standardized list of image criteria and prospectively determined world health organization (WHO) tumor grade and tumor entity. All patients were examined with at least T1- and T2-weighted spin echo sequences before and after contrast injection on 1 and 1.5-T MRI scanners. Following the patients prospectively for 8-13 years after diagnosis, we were able to use nonsurvival at 5 years as a criterion for malignity and reference for the prognostic accuracy of both MRI and tumor tissue histology.

Results: The accuracy for predicting nonsurvival at 5 years was 91% (95% confidence interval (CI): 87-94%) for MRI and 92% (95% CI: 88-95%) for histology. The Kaplan-Meier survival curves of patients with benign and malignant brain tumors as diagnosed by MRI or histology differed significantly (p < 0.001). Histology confirmed benignity or malignity in 201 patients (82%, 95% CI: 77-87%). Sources of misdiagnosis were metastases diagnosed as astrocytoma WHO IV, atypical meningiomas, and low-grade astrocytoma with malignant transformation.

Conclusion: MRI appears as accurate as histology in predicting survival at 5 years after diagnosis. Histological diagnosis may be more specific, however, and is needed to assess the tumor's specific biology.