Structural brain patterns on magnetic resonance imaging (MRI) predict cerebrospinal fluid biomarker levels in patients with suspected Alzheimer's disease (AD) or frontotemporal lobar degeneration (FTLD), preliminary research suggests.
For the study, Corey McMillan (University of Pennsylvania, Philadelphia, USA) and colleagues assessed volumetric MRI data of 32 patients with genetically or autopsy-confirmed AD or FTLD. They used a form of region-of-interest analysis with multivariate adjustment to identify spatial patterns that accounted for the variance in gray matter volumes between the patients.
These patterns predicted patients' ratios of total tau (tt) to β-amyloid (Aβ) and were 75% accurate for distinguishing between patients with AD and FTLD in a larger cohort of 185 patients with diagnoses based on clinical symptoms and lumbar puncture findings.
McMillan et al say that MRI could therefore serve as a screen to identify patients in need of additional diagnostic studies, and could reduce the need for invasive lumbar punctures.
In an editorial accompanying the study in Neurology, Christian Habeck (Columbia University, New York, USA) and Jennifer Whitwell (Mayo Clinic, Rochester, Minnesota, USA) say that the advantage of the team's approach is that "these complex MRI patterns can be summarized with a single diagnostically useful value that could potentially be applied in a clinical setting."
But they stress that the results must be replicated in independent cohorts before the technique can be applied in practice.
Overall, elevated tt/Aβ ratios were associated with reduced gray matter density in posterior areas, including the superior parietal cortex, precuneus, and occipital association cortex. By contrast, lower tt/Aβ ratios were associated with reduced density in frontal regions, including the ventromedial prefrontal cortex, orbital frontal cortex, insula, caudate, and anterior temporal cortex.
In patients with known pathology, actual tt/Aβ ratios were 84% accurate for distinguishing those with AD and FTLD, whereas tt/Aβ ratios predicted by MRI were 75% accurate. These two classification accuracies did not significantly differ from each other.
Among patients with clinical diagnoses, MRI was 75% accurate for distinguishing patients with AD and FTLD when set against actual tt/Aβ ratios.
"Importantly, the clinical diagnoses of the patients with AD and FTLD overlapped substantially, demonstrating utility for predicting pathology in clinically atypical patients in which diagnosis is the most challenging," say Habeck and Whitwell.
They add: "The study also provides an instructive example that could be embraced and copied for other types of diagnostically relevant but hard-to-get physiologic information."
Licensed from medwireNews with permission from Springer Healthcare Ltd. ©Springer Healthcare Ltd. All rights reserved. Neither of these parties endorse or recommend any commercial products, services, or equipment.