Predicting Alzheimer's Disease

Abstract

A great deal of attention is being given to making the diagnosis of dementia and Alzheimer's disease at an earlier point in the clinical spectrum. However, the clinical features of this point on the spectrum are not well delineated. All non-demented subjects from the Mayo Clinic Alzheimer's Disease Patient Registry were eligible for evaluation. Two prediction models were considered: clinical criteria for mild cognitive impairment (MCI) and a cutoff score on the Free and Cued Selective Reminding Test (FCSRT) (<17). Proportional hazards regression models were used to assess the association between each of the two risk indicators and the onset of dementia. Models were fit to each risk indicator individually. An additional model included both indicators simultaneously. Hazard ratios and 95% confidence intervals were obtained from these models. Additionally, C-statistics were calculated to further compare the ability of the two criteria to discriminate between those who progressed to dementia versus those who did not . The significance of the difference between the two C-statistics was assessed using a bootstrap approach. 1,261 non-demented subjects were entered into the evaluation. Of these, 193 progressed to dementia. The mean age and education for the total group were 78.8 years and 13.4 years, respectively. The follow-up ranged from 0.9 to 18.1 years, with a median of 5.2. Clinical MCI status was significantly associated with the outcome of dementia. Those classified as MCI had a hazard ratio for risk of dementia onset of 10.15 (95% CI: 7.59-13.57, p < 0.001). Those with summed learning scores over three trials on the FCSRT of 17 or less were at an increased risk of progression to dementia (HR = 9.10, 95% CI: 6.72-12.33, p < 0.001). When both criteria were included in the same model, both provided independent information for the risk of dementia. The C-statistic calculated for the MCI model (0.78) was 11% higher than that for the FCSRT model (0.70, p < 0.001). Clinical MCI criteria outperformed a single memory test for predicting risk of developing subsequent dementia. Support: Mayo Clinic Alzheimer's Disease Patient Registry UO1 AG06786, P 50 AG16574, the Robert H. and Clarice Smith and Abigail Van Buren Alzheimer's Disease Research Program.