PreMCI, Emci and Amci: Comparison of Imaging Features, APOE Frequencies and Progression Rates to Dementia

Identification of an early predementia stage of Alzheimer's disease as a basis for intervention at a stage when treatment is likely to be most successful, requires definition of a discrete cognitive syndrome that is earlier than MCI, which can be diagnosed reliably. At baseline, subjects were diagnosed with no cognitive impairment (NCI) (n=216), PreMCI (n=125), early MCI (eMCI) (n=90), non-amnestic MCI (naMCI) (n=26), or late MCI (LCMI) (n=30) using the following criteria: (1) PreMCI = CDR score of 0.5, logical memory-delayed (LM-D) test score within 0.5 SD of age and education adjusted means (AEAMs), and all non-amnestic tests within 1.5 SD of AEAMs; (2) eMCI = CDR score of 0.5, MMSE of > 24, LM-D score between 0.5 and 1.5 SD of EAMs; (3) naMCI = CDR score of 0.5, MMSE of > 24, LM-D score within 0.5 SD of AEAMs, and one or more non-amnestic measures at > 1.5 SD below AEAMs; (4) LMCI = CDR score of 0.5, MMSE of > 24, LM-D score > 1.5 SD below AEAMs. APOE allele frequencies, medial temporal atrophy (MTA) scores based on visual ratings of brain MRIs, and progression rates to dementia (on 2-3 year follow-up) were assessed for each subject group. ApoE4 frequencies were not different between groups, although differences in MTA scores were found, i.e., 31% of LMCI, 23% of eMCI, 23% of PreMCI, 4.5% of naMCI, and 9.4% of NCI subjects scored above the threshold (defined as 1.50) (range 0 to 4) [X2 (df=4)=16.88; p < .003]. Differences in rates of progression to dementia were also present, i.e., 23.1% for CDR 0.5 alone; 28% for PreMCI, 34.0% for eMCI, and 43.8% for LMCI [X2 (df = 4)=44.52; p < .001]. Subjects with naMCI or NCI subjects did not progress to dementia. PreMCI and eMCI (but not naMCI) subjects have greater MTA scores and progression rates to dementia than NCI subjects, and are therefore appropriate targets for very early intervention clinical trials. Although the diagnosis of PreMCI is more labour-intensive requiring additional non-amnestic tests, it is likely to be a better-defined and less heterogeneous entity. In the absence of any deficit in LM-D scores, naMCI appears to be a variant of normal cognition.