P1‐511: Towards more strategic subject selection in Alzheimer's randomized clinical trials: neuropsychological testing to distinguish fast decliners from improvers

Alzheimer's disease NINCDS-ADRDA diagnostic criteria require cognitive impairment confirmation by neuropsychological tests in addition to the MMSE or similar examination, yet protocols rarely require documentation of supplemental testing. Broad MMSE inclusion ranges can result in randomization of non-demented subjects and slower than expected placebo group decline, degrading study power. Lopez et al (2010) reported an analysis of pooled data from 14 AD trials, defining criteria for fastest and slowest decline (improvers). We applied these criteria in preliminary analyses to mild/moderate AD study blinded data to identify fast decliners and improvers at week 24; then compared neuropsychological tests at screening. In this phase II trial, MMSE inclusion range was 16-26. At week 24, 69 subjects met Lopez et al Criterion 2: Fast decliners: ADAS-Cog deterioration ≥3.33 points (N = 41), Improvers: ADAS-Cog improvement ≥1.66 points (N = 28). T-tests compared the groups on screening visit tests: MMSE, ADAS-Cog, Fluency, Digit Span, MMSE Attention and Calculation, Color Trails. Distributions were compared to identify potential cut-off scores for trial entry. Preliminary results show statistically significant differences between fast decliners and improvers on Digit Span Backwards, Color Trails 1, MMSE, Category Fluency, Digit Span Backwards plus Attention and Calculation, Color Trails 1 plus 2, ADAS-Cog, Attention and Calculation, Letter Fluency plus Category Fluency, and Color Trails 2. Several neuropsychological tests’ score distributions did not overlap completely (examples: Category Fluency, Digit Span Backwards, Color Trails 1), with worst and best scores uniquely associated with fast decliners and improvers respectively. Measures with restricted or narrow ranges did not have unique values, examples: MMSE and Attention and Calculation. Consistent with the known relationship between severity and rate of decline in AD, in the blinded data fast decliners and improvers were different at screening on verbal fluency, working memory, and complex information processing speed. Further analyses will focus on the smaller placebo group. These measures have potential use in strategic subject selection to fulfill the NINCDS-ADRDA requirement of supplementing the MMSE for confirming cognitive impairment, documenting progressivity (another NINCDS-ADRDA diagnostic criterion neglected in trials), and identifying improvers at screening for exclusion, and should be considered in protocol design.