OBJECTIVE

Increasing evidence suggests that Alzheimer's disease begins at least a decade before the diagnosis of dementia. Earlier identification of the disease will have important implications for intervention; however, current models of preclinical changes are theoretical and require verification from empirical observations. Furthermore, these models have not incorporated psychiatric features.

METHOD

Clinical and biological markers were examined at baseline (1999-2001) in 9,076 people aged 65 years and older. A nested case-control study included 830 cases with Alzheimer's disease diagnosed by DSM-IV criteria during the 10-year follow-up and twice as many controls. By taking the distance between baseline and diagnosis as the length of the preclinical period, disease marker trajectories were estimated using nonparametric locally weighted smoothing analysis.

RESULTS

Significant differences for the cases compared to the controls were observed on both intercept and slope for truncated amyloid β40 (P = .006; P = .003, respectively), C-reactive protein (P = .03; P = .05), verbal fluency (P < .0001; P < .0001), visual recall (P < .0001; P = .007), and hippocampal volume (P = .0002; P = .04) and on the slope only for truncated amyloid β42 (P = .01). The cases showed higher levels of depressive symptoms (P = .003), which remained stable over the 10 years to diagnosis.

CONCLUSIONS

As hypothesized by existing theoretical models, changes in plasma amyloid β levels, hippocampal atrophy, cognitive loss, and C-reactive protein are already observed up to 10 years before diagnosis. An acceleration in cognitive decline appears to follow a significant increase in amyloid accumulation, and depressive symptomatology remains at a constantly higher level. Overall, clinical and biological markers do not follow the same trajectories; the clinical picture changes according to distance from dementia.