Evaluation of a New Criterion for Detecting Prion Disease With Diffusion Magnetic Resonance Imaging.

Importance: Early diagnosis is a requirement for future treatment of prion diseases. Magnetic resonance imaging (MRI) with diffusion-weighted images and improved real-time quaking-induced conversion (RT-QuIC) in cerebrospinal fluid (CSF) have emerged as reliable tests. Objectives: To assess the sensitivity and specificity of diffusion MRI for the diagnosis of sporadic Creutzfeldt-Jakob disease (sCJD) with a new criterion (index test) of at least 1 positive brain region among the cortex of the frontal, parietal, temporal, and occipital lobes; the caudate; the putamen; and the thalamus. Design, Setting, and Participants: This diagnostic study with a prospective and a retrospective arm was performed from January 1, 2003, to October 31, 2018. MRIs were collected from 1387 patients with suspected sCJD consecutively referred to the National Prion Disease Pathology Surveillance Center as part of a consultation service. Intervention: Magnetic resonance imaging. Four neuroradiologists blinded to the diagnosis scored the MRIs of 200 randomly selected patients. One neuroradiologist scored the MRIs of all patients. Main Outcomes and Measures: Sensitivity and specificity of the index test compared with currently used criteria and CSF diagnostic (improved RT-QuIC, 14-3-3, and tau CSF tests). Results: A total of 872 patients matched the inclusion criteria (diffusion MRI and autopsy-confirmed diagnosis), with 619 having sCJD, 102 having other prion diseases, and 151 having nonprion disease. The primary analysis included 200 patients (mean [SD] age, 63.6 [12.9] years; 100 [50.0%] male). Sensitivity of the index test of 4 neuroradiologists was 90% to 95% and superior to sensitivity of current MRI criteria (69%-76%), whereas specificity was 90% to 100% and unchanged. Interrater reliability of the 4 neuroradiologists was high (κ = 0.81), and individual intrarater reliability was excellent (κ ≥0.87). The sensitivity of the index test of 1 neuroradiologist for 770 patients was 92.1% (95% CI, 89.7%-94.1%) and the specificity was 97.4% (95% CI, 93.4%-99.3%) compared with a sensitivity of 69.8% (95% CI, 66.0%-73.4%; P < .001) and a specificity of 98.0% (95% CI, 94.3%-99.6%; P > .99) according to the current criteria. For 88 patients, index test sensitivity (94.9%; 95% CI, 87.5%-98.6%) and specificity (100%; 95% CI, 66.4%-100%) were similar to those of improved RT-QuIC (86.1% [95% CI, 76.5%-92.8%] and 100% [95% CI, 66.4%-100%], respectively). Lower specificities were found for 14-3-3 and tau CSF tests in 452 patients. Conclusions and Relevance: In this study, the diagnostic performance of diffusion MRI with the new criterion was superior to that of current standard criteria and similar to that of improved RT-QuIC. These results may have important clinical implications because MRI is noninvasive and typically prescribed at disease presentation.