[O5–01–04]: SAMPLE SIZES FOR 24‐MONTH ALZHEIMER's PREVENTION TRIALS USING BIOMARKER ENDPOINTS IN COGNITIVELY UNIMPAIRED AMYLOID‐POSITIVE ADULTS

Background: The Severe Impairment Battery (SIB) detects cognitive differences in patients with Alzheimer’s disease (AD) who perform at floor level on other scales. It has six major subscales and three brief evaluations for nine total items; however, adding all item scores may not be optimal for detecting effects in patients with severe AD. A quantitatively-derived SIB composite (qSIB-severe) was used to better detect treatment effects in patients with severe AD, compared with total SIB. Methods:The qSIB-severe was developed using partial least squares (PLS) regression in an independent training dataset comprising pooled data from three 24week, randomized, double-blind, placebo-controlled, memantine studies in moderate-to-severe AD (Reisberg, NEJM 2003; Tariot, JAMA 2004; van Dyck, Alzheimer Dis Assoc Disord, 2007) in which the SIB was administered at weeks 4, 8, 12, 18 (excepting Reisberg), and 24. The qSIB-severe was then tested in severe patients (baseline Mini-Mental State Examination [MMSE] <9) from a separate test dataset (Grossberg, CNS Drugs, 2013), utilizing a mixed effect model with repeated measures. Results:Of 661 patients in the independent test dataset, 142 were severe (mean [SEM] MMSE1⁄46.1 [0.134]); mean (SEM) age was 75.69 (0.702) years. The qSIB-severe included 5 weighted items: language, memory, praxis, attention, and orientation, with a minimum variable importance projection of 0.8101. As early as week 4, memantine least square mean differences (LSMDs) vs placebo were numerically larger with qSIB-severe than total SIB (LSMD [SEM] 1⁄4 0.002 [0.0212] vs -0.650 [2.0143]). Larger LSMDs were also observed at weeks 8 and 12. At weeks 18 and 24, LSMDs were smaller with qSIB-severe than total SIB, though a comparison of effect sizes demonstrated that the qSIB-severe was more sensitive than total SIB at all timepoints (Figure). Conclusions: Compared with total SIB score, the qSIB-severe (language, memory, praxis, attention, and orientation) provided improved measurement sensitivity and differentiation between memantineand placebo-treated patients with severe AD. Quantitatively optimized composite scales, such as the qSIB-severe, may substantially improve sensitivity to treatment effects in subgroups of interest in AD clinical trials, and when utilized early, may inform late-phase study design. O5-01-04 SAMPLE SIZES FOR 24-MONTH ALZHEIMER’S PREVENTION TRIALS USING BIOMARKER ENDPOINTS IN COGNITIVELY UNIMPAIRED AMYLOIDPOSITIVE ADULTS Eric M. Reiman, Xiaoying Kuang, Ji Luo, Yinghua Chen, Pradeep Thiyyagura, Robert Bauer III,, Wendy Lee, Hillary Protas, Dhruman D. Goradia, Paul S. Aisen, Keith Johnson, Jessica B. Langbaum, Reisa A. Sperling, Pierre Tariot, William J. Jagust, Michael Weiner, Kewei Chen, Arizona State University, Tempe, AZ, USA; Arizona Alzheimer’s Consortium, Phoenix, AZ, USA; Banner Alzheimer’s Institute, Phoenix, AZ, USA; University of Arizona, Tucson, AZ, USA; Translational Genomics Research Institute, Phoenix, AZ, USA; Alzheimer’s Therapeutic Research Institute, San Diego, CA, USA; University of Southern California, Los Angeles, CA, USA; Harvard Medical School, Boston, MA, USA; Massachusetts General Hospital, Boston, MA, USA; Center for Alzheimer Research and Treatment, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA, USA; Athinoula A. Martinos Center for Biomedical Imaging, Charlestown, MA, USA; Department of Neurology, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA; Massachusetts General Hospital and the Athinoula A Martinos Center for Biomedical Imaging, Boston, MA, USA; Brigham and Women’s Hospital, Boston, MA, USA; Lawrence Berkeley National Laboratory, Berkeley, CA, USA; University of California, San Francisco, San Francisco, CA, USA; University of California Berkeley, Berkeley, CA, USA; Center for Imaging of Neurodegenerative Diseases, San Francisco Veterans Administration Medical Center, San Francisco, CA, USA. Contact e-mail: eric.reiman@bannerhealth.com