UCSF

Objective

We sought to develop and validate a risk index for prospective cognitive decline in older adults based on blood-derived markers.

Methods

The index was based on 8 markers that have been previously associated with cognitive aging: APOE genotype, plasma β-amyloid 42/40 ratio, telomere length, cystatin C, glucose, C-reactive protein, interleukin-6, and albumin. The outcome was person-specific cognitive slopes (Modified Mini-Mental State Examination) from 11 years of follow-up. A total of 1,445 older adults comprised the development sample. An index based on dichotomized markers was divided into low-, medium-, and high-risk categories; the risk categories were validated with the remaining sample (n = 739) using linear regression. Amyloid was measured on a subsample (n = 865) and was included only in a secondary index.

Results

The risk categories showed significant differences from each other and were predictive of prospective cognitive decline in the validation sample, even after adjustment for age and baseline cognitive score: the low-risk group (24.8%) declined 0.32 points/y (95% confidence interval [CI]: -0.46, -0.19), the medium-risk group (58.7%) declined 0.55 points/y (95% CI: -0.65, 0.45), and the high-risk group (16.6%) declined 0.69 points/y (95% CI: -0.85, -0.54). Using the secondary index, which included β-amyloid 42/40 (validation n = 279), the low-risk group (26.9%) declined 0.20 points/y (95% CI: -0.42, 0.01), the medium-risk group (61.3%) declined 0.55 points/y (95% CI: -0.72, -0.38), and the high-risk group (11.8%) declined 0.83 points/y (95% CI: -1.14, -0.51).

Conclusions

A risk index based on 8 blood-based markers was modestly able to predict cognitive decline over an 11-year follow-up. Further validation in other cohorts is necessary.