Assessment of clinical factors that predict response to nabilone for agitation in Alzheimer’s disease: A post hoc analysis of a randomized placebo-controlled trial

INTRODUCTION: Previously, nabilone showed a medium effect size for treating agitation in moderate-to-severe Alzheimer's disease (AD), but response varied. These post hoc analyses aimed to identify a group of clinical characteristics that predicted treatment response. METHODS: Data from a double-blind, placebo-controlled crossover trial in AD agitation were used. Nineteen clinical characteristics were categorized (presence/absence) and evaluated for relation to agitation response (change on Cohen-Mansfield Agitation Inventory (CMAI)). Characteristics with a ≥ 8 point response difference between categories were included in a multivariable analysis model to calculate individual predicted response. Linear mixed-effects models with Satterthwaite's approximation evaluated the impact of treatment on the relationship between predicted and observed responses. RESULTS: Thirty-nine participants (77 % male, mean [SD] age 87 [10.2], standardized Mini-Mental State Exam (sMMSE) 6.5 [6.8]) were enrolled. Variable selection identified five characteristics related to greater nabilone efficacy: higher pain (Pain Assessment in Advanced Dementia score ≥3) (difference [SE] in CMAI response = -18.8 [3.2]), greater appetite and eating disorders (-16.4 [5.5]), greater apathy (-14.0 [5.5]), less cognitive impairment (sMMSE greater than 10) (-16.5 [4.2]) and no concomitant cholinesterase inhibitors (-13.9 [4.4]). For those with a predicted response in the top tertile based on those five characteristics, 82 % responded, compared with 40 % in the lowest tertile. A treatment-by-tertile interaction (F(2,29) = 8.48, p = 0.001) indicated observed treatment response varied across tertiles. CONCLUSION: A reliable clinical profile of persons with AD related agitation likely to respond to nabilone may be established with additional research.