Purpose: To determine if medial temporal lobe (MTL) atrophy rate, assessed by using an automated procedure over the initial time interval of a 6-year, three-time-point longitudinal study, is predictive of future memory decline.
Materials and methods: Healthy elderly subjects (age, >60 years) were administered a comprehensive battery of neuropsychometric tests and underwent magnetic resonance (MR) imaging at baseline and two or more follow-up examinations. The rate of brain atrophy between the baseline and first follow-up examinations was assessed by using an automated procedure that included spatial coregistration of the two images and regional brain boundary shift analysis. At final observation, the 45 subjects were separated into a group of those who did and a group of those who did not show objective evidence of cognitive decline. A forward stepwise logistic regression model was used to identify variables that predicted decline.
Results: Thirty-two subjects remained healthy, and 13 showed cognitive decline. Among subjects who showed cognitive decline, six declined after the second observation. MTL atrophy rate, through its interactions with sex and age, was the most significant predictor of decline. The overall accuracy of prediction was 89% (in 40 of 45 subjects), with 91% specificity (in 29 of 32 subjects) and 85% sensitivity (in 11 of 13 subjects).
Conclusion: Among healthy elderly individuals, increased MTL atrophy rate appears to be predictive of future memory decline.