Abstract
BACKGROUND: The medial temporal region is the earliest affected structure in patients with Alzheimer's disease (AD), and its atrophy is known as the hallmark of AD. This study aimed to investigate the value of medial temporal atrophy (MTA) for detecting 18F-florbetaben positron emission tomography (PET)-proven AD pathology. METHODS: We retrospectively enrolled 265 subjects complaining of cognitive decline at a dementia outpatient clinic from March 2015 to December 2017. All subjects underwent brain magnetic resonance imaging, 18F-fluorodeoxyglucose PET, and 18F-florbetaben PET at baseline. We performed multivariable logistic regression analyses on variables including age, sex, years of education, white matter hyperintensities, apolipoprotein E (APOE) genotype, and memory composite scores in various combinations to investigate whether MTA was indicative of underlying AD pathology. RESULTS: Our sample population of 265 patients comprised 121 with AD-related cognitive impairment, 42 with Lewy bodies-related cognitive impairment, 32 with vascular cognitive impairment, and 70 with other or undetermined pathologies. In the multivariable logistic regression analyses, MTA was not an independent predictor of underlying AD pathology (P>0.200). The predictive power of underlying AD-related cognitive impairment significantly increased when multiple variables including APOE genotype and memory composite scores were considered together (area under the curve >0.750). CONCLUSION: Our results suggest that MTA alone may be insufficient to accurately predict the presence of AD pathology. It is necessary to comprehensively consider various other factors such as APOE genotype and a detailed memory function to determine whether the patient is at high risk of AD.