Relationship of dementia screening tests with biomarkers of Alzheimer's disease

Abstract

Screening tests for Alzheimer's disease lack sensitivity and specificity. We developed the AD8, a brief dementia screening interview validated against clinical and cognitive evaluations, as an improvement over current screening methods. Because insufficient follow-up has occurred to validate the AD8 against the neuropathologic findings of Alzheimer's disease, we investigated whether AD8 scores correspond to impairment in episodic memory testing and changes in biomarkers of Alzheimer's disease (cerebrospinal fluid and amyloid imaging with Pittsburgh compound B) characteristic of symptomatic Alzheimer's disease. We also compared informant-based assessments with brief performance-based dementia screening measurements such as the Mini Mental State Exam. The sample (n = 257) had a mean age of 75.4 years with 15.1 years of education; 88.7% were Caucasian and 45.5% were male. The sample was divided into two groups based on their AD8 scores: those with a negative dementia screening test (AD8 score 0 or 1, n = 137) and those with a positive dementia screening test (AD8 score ≥2, n = 120). Individuals with positive AD8 scores had abnormal Pittsburgh compound B binding (P < 0.001) and cerebrospinal fluid biomarkers (P < 0.001) compared with individuals with negative AD8 scores. Individuals with positive AD8 tests and positive biomarkers scored in the impaired range on the Wechsler Logical Memory Story A (mean score 7.0 ± 4.5 for Pittsburgh compound B; mean score 7.6 ± 5.3 for cerebrospinal fluid amyloid beta protein 1-42). The AD8 area under the curve for Pittsburgh compound B was 0.737 (95% confidence interval: 0.64-0.83) and for cerebrospinal fluid amyloid beta protein 1-42 was 0.685 (95% confidence interval: 0.60-0.77) suggesting good discrimination. The AD8 had superior sensitivity in detecting early stages of dementia compared with the Mini Mental State Examination. The AD8 had a likelihood ratio of a positive test of 5.8 (95% confidence interval: 5.4-6.3) and likelihood ratio of a negative test of 0.04 (95% confidence interval: 0.03-0.06), increasing the pre-test probability of an individual having symptomatic Alzheimer's disease. Individuals with AD8 scores of ≥2 had a biomarker phenotype consistent with Alzheimer's disease and lower performance on episodic memory tests, supporting a diagnosis of Alzheimer's disease. Informant-based assessments may be superior to performance-based screening measures such as the Mini Mental State Examination in corresponding to underlying Alzheimer's disease pathology, particularly at the earliest stages of decline. The use of a brief test such as the AD8 may improve strategies for detecting dementia in community settings where biomarkers may not be readily available, and may enrich clinical trial recruitment by increasing the likelihood that participants have underlying biomarker abnormalities.

Figure 1

Scatterplots showing CSF Aβ₄₂ levels as a function of in vivo amyloid load as assessed by PiB binding. (A) Subjects are classified by CDR with circles signifying a CDR score of 0 (no dementia) and triangles signifying a CDR score of 0.5 (cognitive impairment). (B) Subjects are classified by AD8 scores with circles signifying negative screening and triangles signifying positive screening. (C) Subjects are classified by consensus clinical diagnoses with circles signifying no dementia, triangles signifying uncertain dementia and squares signifying Alzheimer’s disease. There is excellent correspondence between biomarker profiles and the AD8 similar to the longer CDR and consensus clinical diagnoses. Horizontal reference lines represent the CSF Aβ₄₂ cut-off of 500 pg/ml. Vertical reference lines represent the PiB binding cut-off of 0.18. MCBP = mean cortical binding potential.

Figure 2

Bar graph of scores on the Logical Memory Story A of the Wechsler Memory Scale (WMS) comparing AD8 status (positive versus negative) with PiB status (A) or CSF Aβ₄₂ status (B). Scores for the AD8 negative groups were superior to AD8 positive groups. Mean scores and standard deviations are presented for each group. (A) A two-way ANOVA followed by simple effects tests indicated a significant interaction between AD8 status and PiB binding (P = 0.02) that resulted from the poor performance of those who were positive on both the AD8 and PiB. (B) A similar two-way ANOVA including Aβ₄₂ status instead of PiB produced only a significant main effect of AD8 (P < 0.0001). Neither the main effect of Aβ₄₂ (P = 0.12) nor interaction (P = 0.25) were significant. AB = Aβ₄₂ status.