Diagnostic rates and amyloid positivity vary by definition of mild cognitive impairment

Abstract

AbstractBackgroundThe concept of mild cognitive impairment (MCI) has evolved, and various approaches are used to classify MCI. Conventional MCI criteria, such as those used in the Alzheimer’s Disease Neuroimaging Initiative (ADNI), rely on a single memory test to detect objective cognitive impairment. Actuarial neuropsychological criteria for MCI that incorporate comprehensive neuropsychological performance have been proposed to balance sensitivity and reliability. We applied both actuarial and conventional criteria to a sample of non‐demented older adults. Resulting groups were compared on diagnostic rates and amyloid positivity.Method51 cognitive normal (CN) and 62 MCI participants from the Center for Neurodegeneration and Translational Neuroscience were previously classified by a diagnostic consensus review. Actuarial criteria required impairment (>1 SD below means) on at least 2 tests within a cognitive domain. Conventional criteria were a modified version of ADNI criteria and required Mini Mental State Exam scores ≥ 24 and different cutoffs for Logical Memory based on education. Amyloid positron emission tomography (PET) determined amyloid status.ResultActuarial criteria classified 58 CN and 55 MCI, resulting in 77% agreement with consensus diagnosis. Conventional criteria diagnosed 36 CN and 77 MCI, resulting in 62% agreement with consensus diagnosis. McNemar tests indicated that the proportion of CN and MCI diagnoses were not significantly different between the actuarial and consensus (p=.23) but were significantly different between the conventional and consensus classifications (p=.03), with a 25% false positive rate for the conventional criteria. Chi‐squares demonstrated amyloid positivity was higher in MCI compared to CN for both the actuarial criteria (χ2(1)=31.16, p<.001, Φ=.54) and conventional criteria (χ2(1)=4.71, p=.03, Φ=.21). However, actuarial MCI was twice as likely to be amyloid‐positive than amyloid‐negative (Figure 1), whereas conventional MCI showed similar frequencies of amyloid‐positive and amyloid‐negative status (Figure 2).ConclusionComparisons between MCI diagnostic methods found higher agreement between actuarial criteria and consensus diagnosis versus conventional criteria, which tended to over‐diagnose MCI. A stronger association between amyloid positivity and MCI diagnosis was found using actuarial criteria. Findings support the use of comprehensive neuropsychological criteria to increase the accuracy of identifying individuals with MCI who are at risk for AD dementia.