Alzheimer's disease is thought to be cryptic for several years, with just mild cognitive losses, before a definitive diagnosis can be reached. There has been some evidence that episodic memory (e.g. face recognition) might be more specifically vulnerable in the early stages of the disease. However, a very large retrospective study of Alzheimer's case histories, to come out Neuropsychology tomorrow, shows that patients who later developed Alzheimer's had multiple cognitive losses at stretches up to 10 years before the definitive diagnosis was made:
The analysis showed that no matter what kind of study, people at the preclinical (undiagnosed) stage showed marked preclinical deficits in global cognitive ability, episodic memory, perceptual speed, and executive functioning; along with somewhat smaller deficits in verbal ability, visuospatial skill, and attention. There was no preclinical impairment in primary memory...
..The data also supported the emerging consensus that AD's preclinical period is characterized by an early onset followed by relative stability until a few years before diagnosis, when functioning plummets.
These changes, including a leveling-off of losses, are also seen in normal aging, which is what has made early-stage diagnosis so difficult. (A nice discussion is here )
The study also found that patients who were already under heightened scrutiny for general cognitive losses then showed a stronger episodic memory effect during the early stages; so episodic memory may be an effective metric for disease progression once an at-risk person has been identified. The authors suggest a multi-variable approach to catching the earliest stages of the disease. The huge advantage conferred by early detection is the possibility that treatments will have more effect.
The full article (pdf file) is here .
A nice summary at USA today is here.