Background: Clinical Dementia Rating (CDR) global (CDR-G) and sum of box scores (CDR-SB) are commonly used as primary outcome variables to measure progression or treatment effects in symptomatic Alzheimer disease (AD) clinical trials.

Objectives: We sought to determine whether the CDR is sensitive to change in pre-symptomatic AD and whether there are specific CDR boxes that are dynamic during the multi-year Anti-Amyloid in Asymptomatic Alzheimer's Disease (A4) secondary prevention study.

Design: All participants entered the study with a CDR-G of 0. Box scores were examined individually and as composites of cognition (memory, orientation and judgment /problem solving) and function (community affairs and home/ hobbies). A progression in box score was tabulated only when the change occurred at two consecutive visits.

Setting: The A4 study took place at 67 sites in Australia, Canada, Japan and the United States.

Participants: 1,147 individuals, ages 65-85, were randomized to either placebo (n= 583) or solanezumab (n= 564). All participants received a baseline flobetapir PET scan, an annual CDR, and cognitive testing every 6 months with the Primary Alzheimer Cognitive Composite (PACC) over the course of 240 weeks.

Measurements: Generalized estimating equations and generalized least square models were used to explore the modeled mean progression rate in the CDR-G, CDR-SB, individual CDR boxes, and CDR composite scores in the combined solanezumab and placebo groups. Models were refitted to explore the probability of CDR progression in centiloid tertiles of amyloid at baseline (< 46.1 CL, 46.1 to 77.2 CL, > 77.2 CL). All models included effects for age, education, APOEε4 carrier status, baseline amyloid with flobetapir PET, treatment, and time-by-treatment.

Results: There were no statistical differences between the placebo or solanezumab groups in CDR-G, CDR-SB, specific CDR boxes or CDR composite scores over the course of the trial. Changes in judgment/ problem solving were present at baseline and persisted over time, but progression on the CDR memory box and the CDR cognitive composite quickly predominated. Community affairs and home/ hobbies showed little progression. Personal care remained stable. The probability of cognitive and functional progression in CDR boxes began either at the intermediate or advanced amyloid level (46.1 to 77.2 CL, > 77.2 CL), while amyloid at the lowest level (< 46.1 CL) showed relatively little CDR progression.

Conclusions: The findings suggest that the CDR memory box and the CDR cognitive composite progressed over 240 weeks and were associated with intermediate and advanced stages of amyloid at baseline. Functional changes in community affairs and home/hobbies were relatively stable. These finding suggest that specific CDR box score changes may help refine our measurement of expected treatment effects in future AD prevention trials.