Affiliations: [a] Department of Neurology, David Geffen School of Medicine at UCLA, Los Angeles, CA, USA
| [b] Mallinckrodt Institute of Radiology, Division of Neuroradiology, Washington University, St. Louis, MO, USA
| [c] Department of Neurology, Washington University, St. Louis, MO, USA
| [d] Department of Psychiatry & Biobehavioral Sciences, David Geffen School of Medicine at UCLA, Los Angeles, CA, USA
| [e] V.A. Greater Los Angeles Healthcare System, Los Angeles, CA, USA
Correspondence:
[*]
Correspondence to: Cyrus A. Raji, MD, PhD, Assistant Professor of Radiology and Neurology, Director, Neuromagnetic Resonance Imaging, Division of Neuroradiology, Washington University School of Medicine, Mallinckrodt Institute of Radiology, 4525 Scott Ave., Campus Box 8131, St. Louis, MO 63110, USA. Tel.: +1 314 273 0334; E-mail: [email protected].
Abstract: Background:The differentiation of behavioral variant frontotemporal dementia (bvFTD) from early-onset Alzheimer’s disease (EOAD) by clinical criteria can be inaccurate. The volumetric quantification of clinically available magnetic resonance (MR) brain scans may facilitate early diagnosis of these neurodegenerative dementias. Objective:To determine if volumetric quantification of brain MR imaging can identify persons with bvFTD from EOAD. Methods:3D T1 MR brain scans of 20 persons with bvFTD and 45 with EOAD were compared using Neuroreader to measure subcortical, and lobar volumes, and Volbrain for hippocampal subfields. Analyses included: 1) discriminant analysis with leave one out cross-validation; 2) input of predicted probabilities from this process into a receiver operator characteristic (ROC) analysis; and 3) Automated linear regression to identify predictive regions. Results:Both groups were comparable in age and sex with no statistically significant differences in symptom duration. bvFTD had lower volume percentiles in frontal lobes, thalamus, and putamen. EOAD had lower parietal lobe volumes. ROC analyses showed 99.3% accuracy with Neuroreader percentiles and 80.2% with subfields. The parietal lobe was the most predictive percentile. Although there were differences in hippocampal (particularly left CA2-CA3) subfields, it did not add to the discriminant analysis. Conclusion:Percentiles from an MR based volumetric quantification can help differentiate between bvFTD from EOAD in routine clinical care. Use of hippocampal subfield volumes does not enhance the diagnostic separation of these two early-onset dementias.
