Introduction: We investigated the efficacy and acceptability of pharmacotherapy for dementia with Lewy bodies (DLB) while simultaneously considering the neuropsychiatric symptoms (NPS), cognitive function, motor symptoms, and acceptability.

Methods: Electronic databases were searched from inception through June 5, 2019, for randomized controlled trials (RCTs) and open-label trials (OLTs) in patients with DLB. We performed a pairwise conventional meta-analysis (PWMA) and network meta-analysis (NMA) within a frequentist framework. The main outcomes were mean change scores in NPS, general cognition, motor symptoms and acceptability. The effect sizes and odds ratios with 95% confidence intervals (CIs) were calculated. This study was registered with PROSPERO (CRD42018096996).

Results: In total, we included 29 studies (9 RCTs and 20 OLTs). In the NMA with 9 RCTs, both high- (mean difference [MD] 2.00, 95% CIs, 0.69 to 3.31) and low-dose (1.86, 0.58 to 3.15) donepezil were associated with a greater cognitive improvement than placebo. High-dose zonisamide was associated with greater motor symptom improvement ( -4.10, -7.03 to -1.17]). No medications reached statistical significance regarding improving neuropsychiatric symptoms or developing intolerable adverse effects as compared to placebo. In the second NMA, with 29 studies as an exploratory analysis, aripiprazole and yokukansan may be effective for neuropsychiatric symptoms, while levodopa may be associated with cognitive impairment.

Conclusions: We report the most comprehensive evidence for the selection of pharmacotherapy for treating different clusters of DLB-related symptoms. Due to the limited availability of RCTs on DLB, more well-conducted RCTs are needed for MMA to warrant clinical efficacy in the future.